Social Processes in Children's Learning: Paul Light and Karen Littleton

Social Processes in
Children’s Learning
Paul Light and Karen Littleton
Cambridge University Press

Imitation in Infancy
Jacqueline Nadel and George Butterworth
This is the Wrst book to bring together the extensive modern evidence for
innate imitation in babies. Modern research has shown imitation to be a
natural mechanism of learning and communication which deserves to be
at centre stage in developmental psychology. Yet the very possibility of
imitation in newborn humans has had a controversial history. DeWning
imitation has proved to be far from straightforward and scientiWc evi-
dence for its existence in neonates is only now becoming accepted,
despite more than a century of enquiry. In this book, some of the world’s
foremost researchers on imitation and intellectual development review
evidence for imitation in newborn babies. They discuss the development
of imitation in infancy, in both normal and atypical populations and in
comparison with other primate species, stressing the fundamental im-
portance of imitation in human development, as a foundation of com-
munication and a precursor to symbolic processes.
Learning to Read and Write: A Cross-Linguistic
Perspective
Margaret Harris and Giyoo Hatano
For many years, the development of theories about the way children
learn to read and write was dominated by studies of English-speaking
populations. As we have learned more about the way that children learn
to read and write other scripts – whether they have less regularity in their
grapheme–phoneme correspondences or do not make use of alphabetic
symbols in all – it has become clear that many of the diYculties that
confront children learning to read and write English speciWcally are less
evident, or even non-existent, in other populations. At the same time,
some aspects of learning to read and write are very similar across scripts.
The unique cross-linguistic perspective oVered in this book, including
chapters on Japanese, Greek and the Scandinavian languages as well as
English, shows how the processes of learning to read and spell are
aVected by the characteristics of the writing system that children are
learning to master.
Children’s Understanding of Biology and Health
Michael Siegal and Candida Peterson
This book uses new research and theory to present the Wrst state-of-the-
art account of children’s understanding of biology and health. The
international team of distinguished contributors views children’s under-
standing in these areas to be to some extent adaptive to their well-being
and survival and uses evidence collected through a variety of diVerent
techniques to consider whether young children are capable of basic
theorising and understanding of health and illness. Topics ranging from
babies to the elderly including birth, death, contamination and con-
tagion, food and pain are examined and close links between research and
practice are made with obvious attendant beneWts in terms of education
and communication.
XXXXXX
This page intentionally left blank

Social Processes in Children’s Learning
This book is about children’s learning and problem-solving behaviour.

It reXects the increasingly close integration seen in recent years between
social and cognitive approaches to researching the learning process. In
particular, Paul Light and Karen Littleton examine the ways in which
interactions between children inXuence learning outcomes. They begin
by placing this topic in a broad theoretical and empirical context and go
on to present a substantial series of their own experimental studies,
which focus on children of late primary and early secondary school age.
These investigations address peer facilitation of problem solving, social
comparison eVects on learning and social context eVects upon the
interpretation of tasks. Many of the studies involve computer-based
learning but the Wndings have implications both for classroom practice
and the understanding of the learning process.
This book will be a valuable tool for researchers, teachers and practi-
tioners interested in the social processes of children’s learning.
paul li gh t is Pro-Vice-Chancellor of Bournemouth University, and

has held teaching and research posts in psychology at the Universities of
Cambridge and Southampton. He has published in a wide range of
journals including Quarterly Journal of Experimental Psychology, Journal
of Experimental Child Psychology, Child Development and Journal of Child
Psychology and Psychiatry. His books include The Development of Social
Sensitivity (1979); Social Cognition (with G. Butterworth, 1982);
Children of Social Worlds (with M. Richards, 1986); Learning to Think
(with S. Sheldon and M. Woodhead, 1991); and Context and Cognition
(with G. Butterworth, 1992).
k ar e n l i t tl e to n currently holds an ESRC personal research fellow-

ship at the Open University. She has published in a range of journals
including European Journal of Psychology of Education, Cognition and
Instruction, and Educational Psychology. In addition, she has edited
Learning with Computers (with P. Light, 1999); and Cultural Worlds of
Early Childhood, Learning Relationships in the Classroom and Making Sense
of Social Development (all 1998 and 1999 and all with M. Woodhead &
D. Faulkner).
XXXXXX
Cambridge Studies in Cognitive and Perceptual Development
Series Editors
George Butterworth (General Editor), University of Sussex, UK
Giyoo Hatano, Keio University, Tokyo, Japan
Kurt W. Fischer, Harvard University, USA
Advisory Board
Patricia M. GreenWeld, University of California, Los Angeles, USA
Paul Harris, University of Oxford, UK
Daniel Stern, University of Geneva, Switzerland
Esther Thelen, Indiana University, USA
The aim of this series is to provide a scholarly forum for current theoretical and
empirical issues in cognitive and perceptual development. As the twentieth cen-
tury draws to a close, the Weld is no longer dominated by monolithic theories.
Contemporary explanations build on the combined inXuences of biological,
cultural, contextual and ecological factors in well-deWned research domains. In
the Weld of cognitive devlopment, cultural and situational factors are widely
recognised as inXuencing the emergence and forms of reasoning in children. In
perceptual development, the Weld has moved beyond the oposition of ‘innate’ and
‘acquired’ to suggest a continuous role for perception in the acquistion of knowl-
edge. These approaches and issues will all be reXected in the series which will also
address such important research themes as the indissociable link betweeen per-
ception and action in the developing motor system, the relationship between
preceptual and cognitive development to modern ideas on the development of the
brain, the signiWcance of developmental processes themselves, dynamic systems
theory and contemporary work in the psychodynamic tradition, especially as it
relates to the foundations of self-knowledge.
Titles published in the series
1. Imitation in Infancy
Jacqueline Nadel and George Butterworth
2. Learning to Read and Write: A Cross-Linguistic Perspective

Margaret Harris and Giyoo Hatano
3. Children’s Understanding of Biology and Health

Michael Siegal and Candida Peterson
Forthcoming titles
Nobuo Masataka
The Onset of Language
XXXXXX

Social Processes in Children’s
Learning
Paul Light and Karen Littleton

Bournemouth University and The Open University
PUBLISHED BY CAMBRIDGE UNIVERSITY PRESS (VIRTUAL PUBLISHING)
FOR AND ON BEHALF OF THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE
The Pitt Building, Trumpington Street, Cambridge CB2 IRP
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
http://www.cambridge.org
© Paul Light and Karen Littleton 1999

This edition © Paul Light and Karen Littleton 2003
First published in printed format 1999
A catalogue record for the original printed book is available

from the British Library and from the Library of Congress
Original ISBN 0 521 59308 5 hardback
Original ISBN 0 521 59691 2 paperback
ISBN 0 511 01202 0 virtual (netLibrary Edition)

To Vivienne and Ian
XXXXX

Contents
List of Wgures page x

Preface xiii
Acknowledgements xviii
1 Peer interaction and learning: perspectives and starting points 1

2 Peers and puzzles: a Wrst series of studies 14
3 Computers and learning 27
4 Kings, Crowns and Honeybears: a second series of studies 32
5 Gender agendas 52
6 Social comparison and learning 73
7 Interaction and learning: rethinking the issues 91
References 101
Index 116
ix
Figures
2.1 The Tower of Hanoi page 15

2.2 The computer version of the Tower of Hanoi 19
2.3 The balance beam task 23
4.1 King and Crown software: the map 34
4.2 King and Crown software: the general information
screen 34
4.3 King and Crown software: the pilot information
screen 35
4.4 King and Crown software: the action screen 35
4.5 Honeybears software: the map 44
4.6 The mean level of performance for the ‘pairs’ and
‘individuals’ during each of the three sessions 46
4.7 The mean session 1, 2 and 3 performance of the
‘same’, ‘less’ and ‘more’ able pairings 47
5.1 Success in session 2 by gender 57
5.2 Success in session 2 by session 1 pair type 57
5.3 The mean level of performance for the three diVerent
pair types 60
5.4 The mean level of performance for the girls and boys
on the King and Crown and Honeybears versions of
the task 62
5.5 The mean level of performance for the girls and boys
on the Pirates and Honeybears versions of the task 63
5.6 The mean level of performance for the mixed and
single-sex dyads 66
single-sex pairs for the interacting pairs 68
single-sex pairs for the coacting pairs 69
6.1 The complex rectilinear Wgure 79
6.2 The mean recall score for the girls and boys for the
two modes of task presentation 82
x
Figures xi
6.3 The mean recognition score for the girls and boys for
the two modes of task presentation 83
6.4 The perceptual-motor skills task 85
6.5 The mean ‘hits’ for the girls and boys in the test and
game conditions 86
XXXXXX

Preface
Two decades ago, in a monograph on ‘The Development of Social

Sensitivity’, one of us concluded that: ‘The way is open for much more
detailed and delicate study of the relationship between cognitive develop-
ment and experience in a social environment’ (Light, 1979, p. 117).
Research in the ensuing years has indeed added greatly to our under-
standing of this relationship, and the purpose of the present volume is to
explore one particular aspect of it, namely the relationship between
children’s learning and their experience of interaction with peers.
In common with a great deal of the research undertaken in develop-
mental and social psychology over the last twenty years, our subject
matter can be embraced by the term ‘social-cognition’. However, this
term encompasses a variety of very diVerent research enterprises. On the
one hand, we have research which is concerned with understanding of
social phenomena. This encompasses perception and understanding of
self and other, understanding others’ intentions and emotions, and more
generally the emergence of a ‘theory of mind’. On the other hand, we have
research which examines the ways in which more general aspects of
cognitive development are shaped by social interactions. Here we see
traditional topics of cognitive developmental research such as reasoning
and concept formation analysed in social-interactional terms.
As Butterworth and Light (1982) observed, the relationships between
these various strands of research on socio-cognitive development have
often been poorly deWned and confusing. Butterworth, in that volume,
observed that: ‘theories have been imported from cognitive development
on the one hand and social psychology on the other, to lie in an uneasy
relationship’ (1982, p. 5). The appearance Wfteen years later of a mam-
moth undergraduate text on ‘Developmental Social Psychology’ (Dur-
kin, 1995) reXects the extent to which the rapprochement between devel-
opmental and social psychology has progressed, but Durkin still describes
the diYculties faced by the enterprise as intimidating.
Durkin draws a similar distinction to that drawn above, between social
cognition as concerned with ‘cognition about social phenomena’ and
xiii
xiv Preface
social cognition as concerned with ‘cognition as a product of social interac-

tion’. The present volume addresses social cognition in this second sense,
drawing substantially upon research which we have conducted with vari-
ous European colleagues. More particularly, our focus will be upon social
interactions between learners as an inXuence upon learning.
The capacity for collaborative learning is widely, and perhaps increas-
ingly, seen as a key feature of human cognitive development. For
example, in a major Behavioral and Brain Sciences (BBS) target article,
Tomasello, Kruger and Ratner (1993) address the age old question of
‘what’s so special about humans?’ Their answer is that what is most
distinctive about humans is the possession of a culture. Culture is deWned
in terms of material artefacts, social institutions, behavioural traditions
and languages, all having the capacity to change over time. Their hypoth-
esis is that underlying all this distinctively human paraphernalia of culture
is a fundamental characteristic of human learning, namely a capacity for
socially grounded learning. Such learning is marked by the fact that it
involves not just learning how to do things, but also coming to understand
situations in terms of the purposes and intentions of others, and appro-
priating their points of view upon shared activities.
Tomasello et al. envisage a progression from imitation through instruc-
tion to collaborative learning, the capacity for which is seen as emerging
in the early school years and as being a vital (perhaps the vital) ingredient
in cultural learning. After all, they observe, ‘Human children are born
into a world in which most of the tasks they are expected to master are
collaborative inventions’ (1993, p. 508). They put the spotlight on the
social sensitivity of the learner, attempting to establish the preconditions
for cultural learning at the level of the individual. They ask, in eVect,
‘what do learners have to be capable of to engage in cultural learning?’.
This approach is open to the obvious criticism (taken up by many of the
commentators on the BBS article) that it treats the individual as ‘prior’ to
culture. The counter-view is that cultural learning is itself a product of
culture. After all, opportunities to engage in imitation, instruction and
collaboration are themselves aVorded or constrained by what the culture
oVers, for example by way of apprenticeship or schooling. It may be going
too far to claim that culture creates the child in its own image, but the
relationship is surely at least a bidirectional one.
At some level, the eYcacy of collaborative learning, for example, must
depend upon the individual’s developing psychological model of others.
At the same time, the development of such psychological models must be
dependent upon social experience. We are thus dealing with a dynamic
system of transactions. Disciplinary traditions vary in the ways in which
they conceptualise the relationships involved, but, whether one starts
Preface xv
from the standpoint of the individual developing learner or of the social

and cultural order, it is apparent that social processes mediate learning in
crucial ways.
The role of interaction in learning is an issue of obvious relevance to
education as well as to psychology. The work described in this volume
draws from both literatures. Educational interest in the potential of group
work for fostering children’s learning has a long pedigree, a useful review
of which can be found in Kutnick and Rogers (1994). Educational reports
on both sides of the Atlantic, certainly since the 1920s and 1930s, have
lent encouragement to small group work in the classroom. In the UK, the
Plowden Report of 1967 was in part responsible for a shift from whole
class teaching towards small group teaching at primary school level. The
Report advocated groupwork for a variety of reasons, some to do with
classroom organisation and eYciency, some to do with socialisation, and
some to do with speciWcally cognitive beneWts of this mode of learning.
Today, classrooms in the UK and throughout much of the developed
world typically have children sitting around tables in small groups, rather
than sitting in individual desks in serried ranks, facing the teacher.
The physical arrangement of the classroom can be deceptive, however.
The children sitting in these groups may be working separately at diVer-
ent tasks. Whole class teaching also can and does go on within such
classrooms. Observational classroom research in the UK indicates that
teachers rarely assign tasks that depend upon collaborative modes of
learning, and that rather little in the way of pedagogically eVective group-
work actually occurs (Bennett, Desforges, Cockburn and Wilkinson,
1984; Galton and Williamson, 1992; Tizard, Blatchford, Burke, Far-
quhar and Plewis, 1988). However, there is at least one aspect of the
curriculum in relation to which groupwork has seen something of a
renaissance in recent years, namely the use of computers.
A combination of hardware shortages and a lack of conWdence amongst
many teachers may have conspired to favour a pupil-centred, small group
approach to learning with computers. A survey of UK primary schools in
the mid-1980s found that computers were predominantly used by two or
three children at a time (Jackson, Fletcher and Messer, 1986). Indeed
many of the teachers responding to that survey saw one of the main
educational beneWts of computers as being the fact that they were particu-
larly good at supporting groupwork. With computers it is also often the
case that some of the children in a class have considerable expertise
gained outside school, thus disturbing the usual ‘teacher-centred’ dis-
tribution of expertise (Shrock and Stepp, 1991).
Despite the fact that the typical desktop computer seems to have been
designed for a single user, classroom observations of computer use have
xvi Preface
long emphasised the potential of computers to support eVective learning

interactions (e.g. Cummings, 1985). Relatively autonomous groupwork
around computer resources seems to many observers to oVer opportuni-
ties for genuine discussion between pupils of a kind which is hard to
sustain in other classroom contexts (Scanlon, IssroV and Murphy, 1998).
The social dimensions of children’s learning interactions around com-
puters will emerge as a major focus of this volume.
The role of discussion and disagreement between peers has long exer-
cised researchers concerned with the role of social interaction in
children’s cognitive development and learning. As we shall discuss in
more detail in the Wrst chapter of the book, Piaget has played a key role in
shaping the agenda for developmental psychology in this area. Suspicious
of any suggestion that intellectual development results from ‘trans-
mission’ of knowledge and understanding from experts to novices, Piaget
(e.g. 1932) downplayed the importance of adult–child interchanges in
favour of an emphasis on the productive potential of peer interactions. He
argued that the egocentrism which so limited the thinking of preschool
and early school-age children could be overcome through encounters
with diVerent points of view. DiVerences of points of view between peers
he saw as having particular value as they demand resolution, whereas
diVerences of view between a child and an adult may result simply in
compliance.
Neo-Piagetian developments of these ideas, associated most notably
with the work of Doise and colleagues (e.g. Doise, 1990; Doise and
Mugny, 1984) have bought the concept of ‘socio-cognitive conXict’ to
prominence, and we shall have a good deal to say about it in the pages that
follow. But through other concepts, such as ‘social marking’, Doise
recognised that the social dimensions of learning interactions extend far
beyond the immediate face-to-face encounter. The wider social world of
rules, conventions and social etiquette impinges in a host of more or less
subtle ways on the learning situation, which is thus in an important sense
‘social’ even when no-one else is present.
This wider sense in which intellectual development is a fundamentally
social process was explicitly addressed by Vygotsky, whose rather frag-
mentary writings (e.g. Vygotsky (1931) 1990) have come to be regarded
as seminal in this Weld of research. Vygotsky oVered an account of some of
the mutually adjustive properties which characterise eVective teach-
ing/learning interactions, building on concepts such as the zone of proxi-
mal development. This has helped to foster a tradition of experimental
research on eVective strategies for teaching and learning which has con-
tributed in turn to the development of computer-based intelligent tutor-
ing systems (Wood and Wood, 1996). It has also provided the basis for an
Preface xvii
approach to understanding how peer interaction can facilitate learning

and problem solving. This highlights processes of joint construction of an
appropriate representation of the problem and its solution, most notably
through discussion (Mercer, 1995).
On another level, Vygotsky oVered a starting point for a wider theoreti-
cal development which has come to be known as ‘cultural psychology’
(Crook, 1994; Cole, 1996). Crook uses the biological analogy of a
‘culture’ as a medium in which living material (bacteria, tissue cells,
or whatever) can be supported and ‘grown’. In a similar way, human
culture is conceptualised as a medium which supports the development of
thinking.
Cultural psychology focuses on thinking and reasoning as activities
which take place in particular situations. This approach has come to-
gether with inXuences from anthropology (Lave and Wenger, 1991) and
cognitive science (Suchman, 1987) to form the basis of what is now
widely referred to as the ‘situated learning’ approach. Here learning is
considered to be an intrinsic and inseparable aspect of participation in the
various ‘communities of practice’ that make up a society.
All of these very diverse disciplinary, sub-disciplinary and multi-disci-
plinary perspectives see development and learning as dependent in one
way or another upon the nexus of social relationships within which they
occur. Many of them suggest that interactions between learners, as well as
between teachers and learners, may have an important and formative
role. In the chapters that follow, some of these perspectives will be
explored in greater detail than others. The research evidence to be pres-
ented at times may lend itself to explanation in terms of one approach
rather than another. But the diVerent positions sketched here are not in
conXict, or at least not in the sense that if one is right the others are
necessarily wrong. All of them oVer a case for taking child–child interac-
tion seriously in the context of development and learning. The present
volume aVords us the opportunity both to pull together a considerable
body of empirical research on this topic that we and our colleagues have
been directly involved in, and also to set this work within the wider
context of contemporary research in the Weld.
Acknowledgements
The writing of this volume was made possible by Economic and Social
Research Council Fellowship awards to both authors (Award numbers:
H52427501595 and H52427000994). Carole Kershaw and Lynda Pres-
ton gave invaluable help in preparing the manuscript for publication. Ian
Wallhead and Simeon Yates kindly reproduced the Wgures. The empirical
research on which the volume is based was supported by project grants
from the ESRC, the Leverhulme Trust and the NuYeld Foundation.
Other members of the research teams included Martin Glachan, Chris-
topher Colbourn, Richard Joiner, Peter Barnes, David Messer, Agnes
Blaye, Maria Silvia Barbieri and Annerieke Oosterwegel. The research
also involved many schools and many hundreds of children, without
whose willing and unrewarded cooperation none of the research could
have been conducted. Our sincere thanks to them all.
Figures 4.1 and 4.2 are reprinted with permission from Springer-Verlag.
These Wgures originally appeared in Computer Supported Collaborative
Learning (1995), edited by Claire O’Malley.
Figure 4.5 is reprinted with permission from Carfax Publishing Limited.

This Wgure originally appeared in Educational Psychology (1998), Volume
18, Number 3.
xviii
XXXXXX

1 Peer interaction and learning: perspectives
and starting points
Introduction
This chapter will set the scene for those that follow by going back to the
1970s to examine some of the strands of theory and empirical research
which converged around the question of when and how peer interaction
can facilitate children’s understanding and learning. Starting with social
learning theory, we shall develop a focus on the concept of socio-cognitive
conXict as an engine of mental development, a concept that owes its
origins to Piaget, via Doise and colleagues. The Piagetian origins of this
idea are reXected in the selection of tasks in these early studies, and this in
turn gives rise to certain problems in terms of the interpretation of some
of the experimental Wndings. This consideration will take us on a slight
detour in the course of this chapter, raising issues to be returned to later.
The latter part of the chapter will be given over to an account of the series
of experimental studies of peer facilitation of children’s problem solving
which marked our own initial engagement with this Weld of research.
Modelling success: the social learning theory approach

As inheritors of the behaviourist approach to learning, social learning
theories exerted a strong inXuence upon the psychology of child develop-
ment in the 1970s. Such theories saw modelling as a key formative process
in cognitive as well as other aspects of development. Thus any facilitative
eVects of child–child interaction in learning were construed largely in
terms of processes of imitation or modelling.
Studies were conducted in which children’s performance on various
cognitive tasks was assessed before and after they had watched other
children performing the same tasks. Many of the tasks in question were
drawn from the Piagetian repertoire, because social learning theorists
were setting themselves against Piagetian constructivist explanations
which they saw as unduly individualistic in emphasis (e.g. Murray, 1974;
Rosenthal and Zimmerman, 1972). Such studies did provide some
1
2 Social processes in children’s learning
evidence that children who initially performed at ‘pre-operational’ levels

on tasks such as conservation of quantity could be induced to give
operational judgements on such tasks simply by being required to observe
another child who oVered such judgements. From a Piagetian point of
view, however, such demonstrations were unconvincing since they did
not oVer evidence that the children concerned were able either to justify
or to generalise these ‘operational’ judgements.
Other studies within this tradition went beyond passive observation to
examine the eVects of actual interaction between ‘pre-operational’ and
‘operational’ children in the context of such tasks (e.g. Miller and Brow-
nell, 1975; Silverman and Geiringer, 1973). Here again, it was possible to
show that the pre-operational children did tend to make progress, and in
this case with some evidence that the children concerned could produce
their own justiWcations for their new operational judgements.
However, in the interaction studies, it was harder to tie such progress
to the child’s exposure to a partner who ‘modelled’ the correct con-
clusion. A diVerent kind of interpretation was developed in relation to
essentially similar studies by researchers working within a ‘constructivist’
tradition.
Construction of understanding through

socio-cognitive conXict
Piaget’s own early writings (most notably, Piaget, 1932) oVered an argu-
ment for the potential productivity of peer interaction in relation to
cognitive development, and especially in relation to the achievement of
concrete operational modes of thought in the early school years. Piaget
saw the pre-school child’s egocentrism as presenting the major obstacle to
achievement of operational thinking. Such thinking required ‘decentra-
tion’, the ability to take account of multiple points of view, and more
generally, multiple covarying factors in a given situation. Pre-schoolers
tended to Wx on the Wrst relevant factor they identiWed, and to answer
entirely in terms of that. What the child needed in order to progress was
something which disturbed this ‘centration’. Exposure to someone else
who saw things diVerently, in a situation which called for resolution of the
conXicting responses, was seen as providing just this kind of disturbance.
Importantly, Piaget considered that inequalities of power and status
were inimical to the eVectiveness of this process. If children were exposed
to the response of a powerful Wgure such as an adult, they would be
unlikely to take issue with it. Rather, they would tend to ignore it if
possible, and comply with it if not. In the case of exposure to a diVering
point of view from an equal, however, the social dynamics of the situation
Peer interaction and learning 3
would create a pressure towards resolution of diVerences: ‘Criticism is

born of discussion and discussion is only possible amongst equals’
(Piaget, 1932, p. 409). Even if the second child’s answer was as wrong as
that of the Wrst, the attempt to resolve their partial and ‘centred’ solutions
would be likely to result in the achievement of a higher level, more
decentred representation which could embrace what was correct in both
of the initial oVerings.
In the mid-1970s, Doise and colleagues in Geneva conducted a series
of experiments on the eVects of peer interaction on the transition to
operational modes of thinking in Wve to seven-year-olds (Doise, Mugny,
and Perret-Clermont, 1975, 1976; Mugny and Doise, 1978). These
studies used a variety of Piagetian ‘concrete operations’ tasks, a favourite
being a ‘village’ task based loosely on Piaget’s famous ‘three mountains’
task (Piaget and Inhelder, 1956). Here, model buildings were arranged
on a baseboard to form a little village. The buildings are oriented in
relation to a Wxed mark on the baseboard, depicting, say, the village pond.
The whole arrangement sat on a tabletop in front of the child. To the side
of the child was another table, with an identical baseboard, but perhaps
oriented diVerently in relation to the child. The child’s task was to use a
replica set of model buildings to recreate exactly the same village on this
second table. The task is more or less diYcult depending upon the
relative orientations of the two baseboards. Where a rotation relative to
self is involved, pre-operational children will typically fail to take account
of the reorientations necessary to preserve the relationships between the
buildings and the Wxed mark.
Children were Wrst tested individually on the task, to get a ‘pre-test’
measure of performance. In a second session, perhaps a week later,
children were given another opportunity to do similar tasks, but this time
some of them worked in pairs or small groups while others worked alone.
Assignment to conditions was essentially random, although in some
studies allocations to particular groupings was on the basis of pre-test
performance. Sometimes, when the ‘village’ task was used, the children
were put in diVerent positions relative to the tables, so that the necessary
transformation was easier for one child than for the other. In a third
session, a week later again, all of the children were given a post-test
individually.
Through such studies, Doise and his colleagues were able to show that
children of slightly diVerent pre-test levels, working together in dyads or
triads, tended to perform at a higher level when working as a group than
children working alone. More importantly, this beneWt carried over to the
children’s individual post-test performances. In other words, the extent
of pre- to post-test progress was signiWcantly greater for those children
who had worked in pairs in the second session than for those who had
worked alone.
Large diVerences between the children in terms of their pre-test levels
were associated with less progress than small diVerences in initial ability
(Doise and Mugny, 1984). Even children who showed identical levels of
pre-test performance could beneWt from working together if steps were
taken to ensure that they would come up with conXicting responses.
Thus, with the village task, Doise and colleagues arranged for such
children to occupy diVerent spatial positions relative to the array. This
meant that their ‘egocentric’ responses would ensure that they came into
conXict about where to place the buildings, even though they were both
reasoning in the same way. Children paired under these conditions made
more gains than similar children working on the tasks on their own.
Doise and colleagues interpreted their Wndings in terms of socio-
cognitive conXict. The children in the pair or small group conditions
found themselves confronted with solutions which conXicted with their
own. This conXict, and the socially engendered need to resolve it,
prompts the children to re-examine their own initial responses, and may
lead them to recognise a higher order solution to the problem which
resolves the apparent conXict (Mugny, Perret-Clermont and Doise,
1981). For this to occur, it is necessary that the children’s initial solutions
diVer, but it is not necessary for any of them to be more advanced than the
others, or for any of them to be correct. The real ‘ratchet’ driving the
process is that resolution of children’s partial or centred solutions can in
the end only be found by adopting a higher level, more decentred sol-
ution, thus ensuring cognitive progress.
Perret-Clermont (1980) used essentially the same three-stage experi-
mental design in a series of experiments on peer facilitation of conserva-
tion judgements. Pre-tests were individual, and involved a range of stan-
dard Piagetian assessments of children’s understanding of conservation
of quantity. Various arrangements were tried out for the children assigned
to the social interaction condition in the second session. One which
proved eVective was to assign two ‘conservers’ and one ‘non-conserver’ to
work together. The non-conserver was then given a task such as sharing
out between the children by pouring from a jug into three diVerent
shaped glasses. The session went on until all the children agreed that they
each had the same quantity. The non-conserving children exposed to this
kind of interaction went on to show signiWcant pre- to post-test gains on
standard tests of conservation of liquid quantity.
With conservation, as with the village task, progress could result even
from interaction between two non-conservers, provided that they gener-
ated diVering initial judgements. These studies further highlighted the
potential beneWts of peer interaction for the development of children’s

thinking, and the supposedly key role of socio-cognitive conXict in
underpinning them. They certainly helped to stimulate research interest
in this area, and indeed triggered our own initial studies reported later
in this chapter. Not surprisingly, however, they also engendered some
controversy.
Reservations about socio-cognitive conXict

Doise and colleagues attracted a good deal of attention in the early 1980s,
and not a little criticism. Russell (1981, 1982) argued that in tasks such as
conservation the pre-operational child was in eVect responding with an
opinion of how the array looked after transformation. By contrast, the
operational child was responding in terms of what is objectively the case.
The diVerence, as Russell sees it, is one of ‘propositional attitude’.
Pre-operational children coming up with diVering answers could quite
well (and, Russell suggests, often do) simply agree to disagree. If and
when conXict is eVective, on this argument, it should be by prompting the
children towards the adoption of an appropriate objective attitude, allow-
ing them to bring to bear understandings that they already possess.
The idea that young children might in fact understand more about
conservation than their responses on standard Piagetian tests would
suggest was one of the main themes of Donaldson’s inXuential book
Children’s Minds (1978). McGarrigle and Donaldson, as early as 1975,
were able to show that manipulations to the context of presentation of
conservation tasks, such as having the transformation of materials appear
accidental, could have a dramatic eVect upon children’s responses.
Donaldson suggested that the standard conservation procedure con-
tained misleading socio-communicative cues to the child to attend to
appearances, rather than to the actual quantities concerned. The very
deliberate way in which the transformation of apparent quantity was
eVected made this the natural focus for the child’s attention. When the
transformation occurred as a seemingly accidental consequence of the
activities of an errant teddy bear, the child was able to discount the
transformation as irrelevant to the actual quantities involved.
Other similar studies followed. In some studies of our own, we were
able to show that even where the transformation of materials was deliber-
ately done by the experimenter, children would discount it if some
plausible rationale for it was provided. Thus Light, Buckingham and
Robbins (1979) used a badly chipped beaker as a reason for pouring the
contents from one container to another, which just ‘happened’ to be of a
diVerent shape. This was done in the context of setting up a game for the
children, who were tested in pairs. The game was such that it was
important for quantities to be equal, but the chip in the beaker would
make the game dangerous; thus the need to Wnd another container. A
substantial majority of Wve and six-year-olds judged that the resulting
transformation did not aVect the quantities involved, whereas almost all
children of this age fail on the standard version of the same task.
This result has proved replicable (e.g. Miller, 1982), but the interpreta-
tion put on it at the time may well be the wrong one. That interpretation
stressed the fact that the transformation was made to seem incidental to
the proceedings, rather than central to them. Later research using diVer-
ent designs (Roazzi and Bryant, 1992) has failed to Wnd signiWcant eVects
for such ‘incidentality’. On the other hand, our own subsequent studies
have shown that setting the conservation task in the context of a game is
suYcient by itself to improve children’s performance.
For example, Light, Gorsuch and Newman (1987) presented pairs of
Wve and six-year-olds with a heap of dried peas for them to divide into two
equal heaps. These were then put into two rather diVerently shaped
containers, and the children were asked whether the quantities remained
equal. For some of them, all this was done in the context of setting up a
game in which they were going to compete with one another to move their
peas as fast as they could to a target container, using a straw. The children
who encountered the conservation questions in this context were much
more likely to respond correctly than those who encountered the same
transformation and the same questions without the game setting. So it
seems that when children are working in pairs and anticipating a competi-
tive game, they construe the conservation test procedures diVerently. In
this situation they remain resolutely attentive to the quantities involved,
and are not readily distracted by the appearances.
Things are actually rather more complicated. It can be argued (Light,
1986) that in these modiWed versions of the conservation test the children
are really just complying with the apparent wishes and expectations of the
experimenter. Just as the standard versions of these tasks might lead
children toward the wrong response, these versions may in various ways
cue the correct response, even in children with no Wrm grasp of conserva-
tion. Indeed we have been able to show that similar modiWcations can
elicit ‘conserving’ responses even in situations where the quantity in
question is not in fact conserved (Light and Gilmour, 1983).
For present purposes, however, this is not really important. What is
important is that serious question marks were appearing about the valid-
ity of these kinds of Piagetian tests, and the stability of children’s re-
sponses to them. More particularly, there are clear suggestions from this
literature that children in pairs or small groups may well interpret given
tasks and questions diVerently from the way in which the same tasks and
questions would be understood by individuals. Improvements in per-
formance, at the time and subsequently, may reXect this altered under-
standing of the questions at least as much as it reXects any socio-cognitive
conXict arising from diVerent points of view within the group.
One thing that makes this alternative point of view attractive is that
peer facilitation processes sometimes seem to work just too well. Thus, for
example, Perret-Clermont (1980) found that social class diVerences in
children’s pre-test performance on conservation tasks were typically
large, but after a session of interaction of the kind described earlier, the
diVerences according to class disappeared. Similar Wndings have been
reported for urban–rural diVerences, and ethnic diVerences (Light and
Perret-Clermont, 1991).
If the pre-test diVerences were genuinely reXective of diVerences in the
children’s levels of achievement in this crucial area of cognitive develop-
ment, is it really conceivable that such diVerences could be wiped out by a
single session of ten to Wfteen minutes of interaction? It seems much more
plausible that the initial diVerences reXect diVerences in children’s under-
standing of the meaning and reference of the conservation questions, and
indeed there is some independent evidence that this is the case (Grossen,
1988).
It would seem that what disambiguates the questions for the children is
the experience of sharing in the paired or group session, leading Light and
Perret-Clermont to conclude that: ‘the eYcacy of the peer interaction
procedure arises not (or not only) from the socio-cognitive conXict mech-
anism . . . but from the introduction of a norm of equality which serves to
support correct responses, which are then carried over to the individual
post-test’ (1991, p.145).
This issue of how social norms inXuence cognitive functioning surfaced
in a number of areas of research in the 1980s, not least in Doise’s own
work on the inXuence of ‘social marking’. However, we shall leave the
further exploration of this issue for a later chapter. In the remaining
sections of this chapter, we shall turn from Piaget to Vygotsky to Wnd a
rather diVerent set of ‘starting points’ for research on social (and more
particularly peer) processes in learning.
Learning as the co-construction of understanding

Vygotsky’s writings of the 1920s and 1930s, though they led to a robust
tradition of research in the Soviet Union, had little real impact on West-
ern developmental psychology for nearly half a century. In the 1980s,
however, there was a rush of translation, commentary and exploitation of
‘Vygotskian’ approaches in relation to a whole range of research problems

(e.g. Bruner, 1985; Wertsch, 1985; Newman, GriYn and Cole, 1989;
Forman, Minick and Stone, 1993).
Vygotsky’s work has perhaps contributed to this Weld in two main ways.
The Wrst rests on his attempts to characterise the Wne-grained interper-
sonal interactions that take place in learning settings, and involves con-
cepts such as the ‘zone of proximal development’ and ‘scaVolding’. The
second rests on his broader attempt to develop a ‘cultural psychology’,
within which learning is seen to depend upon mediation by social, cul-
tural and institutional processes at many levels. We shall explore both of
these contributions.
Vygotsky was interested in the origins of what he termed the higher
mental functions: thinking, reasoning and understanding. The develop-
ment of these higher mental functions in humans was seen as a funda-
mentally social rather than individual process. The child’s interactions
with other people serve to mediate between the child and the world-to-
be-learned-about, and so understanding learning depends upon under-
standing the particular types of interactions which serve to foster it.
The concept of a zone of proximal development is central to this
approach. Children, or indeed adults, can be characterised in terms of
what they can achieve unaided. Indeed most forms of assessment involve
testing what individuals can do without help. But individuals may also
diVer in terms of what they can do with help. The attainments which are
possible for an individual given a measure of support and guidance are, as
Vygotsky put it, within that individual’s zone of proximal development
(ZPD). They are attainments that will be possible for that individual
unaided at some point in the near future.
The concept of a ZPD is thus an integral part of a theory of teaching
and learning. Tharp and Gallimore (1988) have elaborated this aspect of
Vygotsky’s work into what they call a theory of teaching as assisted
performance. They see learning as a process of guided re-invention,
whereby social guidance makes it possible for the learner to achieve a
constructive intellectual ‘re-invention’ of some piece of culturally elabor-
ated knowledge. The emphasis upon understanding being a matter of
construction is clearly shared with Piagetian approaches. The distinctive
features are (i) recognition that much of what the learner needs to learn is
already in some sense ‘available’ in the culture, and (ii) recognition that
interpersonal processes play a key role in making that culturally elabor-
ated learning available to the individual.
Not unnaturally, given this emphasis on guidance, Vygotsky saw the
relevant interpersonal interactions as going on between the learner and a
more capable ‘other’. Indeed, he deWned the ZPD as: ‘The distance
between the actual developmental level as determined by individual

problem solving and the level of potential development as determined
through problem solving under adult guidance or in collaboration with more
capable peers’ (Vygotsky, 1978, p. 86, our emphasis). As with Piaget, then,
peer interaction is Xagged as having a potentially important role in learn-
ing and development. But whereas Piaget’s emphasis was on the status-
symmetry of such interactions, Vygotsky’s emphasis is more on the
competence-asymmetry that will often be a feature of peer relations. As
Tharp and Gallimore put it, to the extent that peers can assist perform-
ance, learning will occur through their assistance.
The concept which has been most widely used to capture the forms of
guidance which support learners in their progress through the ZPD is that
of ‘scaVolding’. Introduced by Wood, Bruner and Ross (1976), it captures
the sense in which, through encouragement, focusing, demonstrations,
reminders and suggestions, a learner can be supported in mastering a task
or achieving an understanding. To take the building analogy further, if we
imagine building an arch with bricks it is easy to see the vital role played by
the wooden ‘formwork’ used to assemble the arch. However, the role of
this scaVolding is strictly temporary; when complete the arch will hold
itself up, though without scaVolding it could not have been built.
Tharp and Gallimore see progression through the ZPD in terms of four
stages. In the Wrst, performance is directly assisted by more capable others
through ‘scaVolding’ of one kind or another. In the second, the learner
eVectively takes over the role of the ‘scaVolder’ in relation to his or her
own learning. This often means ‘talking oneself through’ a task, remem-
bering requests, reminders and injunctions previously given, and so on.
The third stage is marked by the falling away of such ‘self-guidance’, as
performance becomes automatic. The fourth ‘stage’ just recognises the
fact that we can get thrown back to earlier stages of the acquisition
process by such stressors as tiredness, or by changes in the precise
conditions of the task. Learning to drive provides a useful case in point for
all of these stages of learning.
It is as a model of adult guidance of children’s learning that most direct
use has been made of these concepts. For example, Wood and colleagues
(see Wood, 1986) conducted a series of investigations of how four-year-
old children can be taught to assemble a 3D puzzle involving wooden
blocks and pegs. First they observed mothers’ attempts to teach their own
children how to complete the puzzle. The mothers who succeeded best
were those who shifted their levels of intervention Xexibly according to
how well the child was doing. This ‘contingency strategy’ can be seen as a
way for the mother to gauge and monitor the child’s ZPD as learning
proceeds, and to provide scaVolding at just the right point.
In further studies Wood and colleagues showed that the adoption of a

‘contingent’ strategy by specially trained tutors also resulted in better
learning outcomes than any of the alternatives explored. In their recent
work (Wood and Wood, 1996, in press), they have been working towards
the development of computer-based tutoring systems which will provide
optimally contingent patterns of tutorial support; something which hu-
man tutors, even given training, Wnd it extremely diYcult to do.
Although Wood and colleagues are intent on improving upon ‘what
comes naturally’, much of the research done from a Vygotskian stand-
point has tended to see eVective teaching and learning exchanges as
essentially incidental to ongoing joint engagement in activities, whether
between mothers and children in the home (e.g. RogoV and Gardner,
1984; RogoV, 1990) or between experts and novices in traditional craft
practices such as tailoring or weaving (GreenWeld and Lave, 1982; Lave
and Wenger, 1991).
Attempts have been made, however, to use a Vygotskian approach to
illuminate what is going on in the classroom. Perhaps the most successful
is that of Edwards and Mercer (1987), who explored how, by skillfully
guiding classroom discussion, the class teacher can establish and main-
tain a focus of shared attention, provide children with a language in
which to describe their own experiences and, using that language, build
up a body of ‘common knowledge’ about the topic in hand. On the other
hand, attempts to apply the concept of scaVolding more directly in
relation to classroom practice have in at least some cases (e.g. Bliss,
Askew and Macrae 1996) been unable to Wnd any evidence for such
processes at all.
Vygotsky’s ideas about the social-interactional bases of learning have
also inspired a considerable number of studies of learning in the context
of peer interaction. As one might expect, a good proportion of these
involve interactions in which an older, more experienced or pre-trained
individual is designated as tutor to a younger, less experienced or un-
trained peer. This kind of ‘peer tutoring’ has been shown to be eVective
both in experimental studies (e.g. Phelps and Damon, 1989) and in
applied educational contexts (e.g. Topping, 1994). It may indeed have
beneWts for the peer tutors as well as the peer tutees (Barron and Foot,
1991).
As Hogan and Tudge (in press) note, there has been relatively little
research on peer collaboration, as opposed to peer tutoring, approached
from a Vygotskian standpoint. However, increasingly since the mid
1980s, Vygotskian ideas have come to colour the interpretations oVered
by researchers for peer facilitation eVects observed in experimental
studies.
For example, Forman and Cazden (1985) studied joint problem solv-
ing by pairs of nine- to fourteen-year-olds. Their problem-solving tasks
included Piaget’s chemical combinations task (Inhelder and Piaget,
1958) and a task involving predicting the shape of the shadows of vari-
ous objects, when lit from various positions. Their observations of ex-
changes between the children highlighted the importance of well co-
ordinated joint activity in, for example, setting up the apparatus and
planning the experiments. Establishing shared goals and building a
shared understanding of the task are just the kinds of ‘co-constructive’
activities that a Vygotskian analysis might lead one to look for in eVec-
tive peer interaction. On the other hand, in the interpretation of the
results of the experiments, Forman and Cazden saw clear indications
that diVerences of interpretation, defended and argued over, could be
productive in much the way that a Piagetian analysis would lead one
to expect.
Both theoretical approaches may thus have something to oVer, in
drawing attention to diVerent aspects of productive interaction. While a
Piagetian approach highlights what being confronted with a socio-cogni-
tive conXict can do for individual cognitive development, a Vygotskian
approach highlights the way in which a shared understanding can be
arrived at through a process which might be termed ‘mutual construc-
tion’. We shall return to this issue in more detail in subsequent chapters,
but here we want to turn to a rather wider reading of what Vygotsky might
have to oVer to an understanding of social processes in learning.
A socio-cultural approach to learning

The study of social processes in learning can lead into polarised arguments
about ‘which comes Wrst, the individual or the social/cultural setting?’.
Vygotsky attempted to avoid either kind of reduction by focusing on
‘mediated action’, adopting a broad theoretical approach to development
which is nowadays usually referred to as a socio-cultural approach.
Wertsch (1997, 1998) summarises the properties of mediated action in
the following terms. Mediated action involves a relationship between an
agent and a ‘mediational means’ or a ‘cultural tool’. Wertsch uses the
example of pole vaulting: both the pole (a culturally given tool) and the
pole vaulter are intrinsic to the activity. A more cognitive example might
be the use of a word-processing or drawing package on a computer; the
cultural tool both lends itself to being used in various kinds of ways and at
the same time imposes various kinds of constraints.
The notion of cultural tools extends to symbolic tools elaborated
within a culture, so that a mathematical algorithm which allows you to
do mental arithmetic is just as much a cultural tool as is a pocket

calculator. Language itself (or at least any particular language, elabor-
ated for any particular set of purposes) can be considered as a cultural
tool. Thus virtually all intelligent activity involves interacting with a
range of cultural tools, and competence in the use of such tools is central
both to intellectual development and to becoming an eVective member
of society.
To return to Wertsch’s example of the pole vaulter, the advent of a new
type of pole can make new records possible. Probably some pole vaulters
will adopt the new type of pole with enthusiasm and ‘make it their own’,
while others will stick resolutely with the old technology. This sense of
‘making something your own’ is often referred to as ‘appropriation’, and
catches at the sense in which cultural tools are not just picked up and put
down as and when needed. Rather they become part of how we construe
the world, how we approach problems and even how we relate to one
another.
The appropriation of cultural tools involves more than simply having
access to them. Cultural tools in general are associated in complex ways
with the distribution of power and authority within the culture. In the
case of computers, for example, powerful vested interests shape the
resources which become available to individuals, and issues of access and
equity arise very pointedly (Light, 1997; Littleton, 1995, 1996).
A socio-cultural approach thus directs us not only to look at the
potentialities of collaborative interactions for learning in a particular way,
but also to relate the interactional processes observed to the institutional
contexts in which such collaborative interactions occur. This consider-
ation will come to be a salient one in relation to some of the issues such as
gender which will occupy us in later chapters in this volume.
Piaget and Vygotsky: complementary constructions

of collaboration
What becomes evident from this review of Piagetian and Vygotskian
approaches is that they have many features in common which diVerenti-
ate them from learning theory traditions. For both, learning is a matter of
active construction, the ingredients for which are to be found in the
physical and social world. Vygotsky’s stress on the mediated character of
action extends Piaget’s account in ways which add to its reach in import-
ant respects, while not quarrelling with its basic tenets. Between them,
these two approaches to understanding development and learning have
shaped almost all of the research to be described and discussed in this
volume. We shall have occasion to reXect at various points on how well
these perspectives have served their purpose in framing this Weld of

research.
In the next chapter, though, we shall move rather sharply from the
general to the particular, by giving a synopsis of the series of experimental
studies which represent our own Wrst attempts to come to grips with the
phenomena of collaborative learning.
2 Peers and puzzles: a Wrst series of studies
Introduction
Our own interest in conducting empirical research on peer interaction as
a facilitator of children’s problem solving was raised mainly by exposure
to the neo-Piagetian research of Doise and colleagues, described in the
previous chapter. However, the doubts raised earlier about the validity of
Piagetian procedures for assessing conservation were also being raised by
our own and others’ research in relation to a wide range of other Piagetian
tests. These included the spatial perspective taking tasks which had been
so extensively used by Doise and colleagues in their peer interaction
studies (Donaldson, 1978; Light and Nix, 1983).
Since our interest was the more general question of when, if at all, ‘two
heads are better than one’, it seemed sensible to move beyond the
particular case of children’s mastery of ‘concrete operational reasoning’
to consider a wider range of ages and types of task. Nonetheless, as will be
apparent from the account which follows, the basic three-stage experi-
mental design which had been adopted by Doise and others was carried
over fairly intact into these studies.
The Wrst Tower of Hanoi study

The Tower of Hanoi is a traditional game which has not infrequently
been used by cognitive psychologists to study problem solving. A typical
version is shown in Figure 2.1. Here the materials consist of a baseboard
with three vertical pegs, and a number of ‘tiles’. These each have a hole in
and can be placed on any of the pegs (the little handles on the tiles are an
adaptation for our particular purposes). Figure 2.1 shows a possible
initial position, with all the tiles being stacked in order of size on peg (A).
The object, on any particular occasion, might be to move all the tiles to
peg (C), so that they end up similarly ordered by size on that peg. The
constraints are that only one tile may be removed from a peg at a time,
and at no point can a larger tile be placed on top of a smaller one. If three
14
Peers and puzzles 15
Figure 2.1 The Tower of Hanoi
tiles are to be moved, as in Figure 2.1, then the player has to move the
smallest to (C), the middle tile to (B), the smallest to (B), the largest to
(C), the smallest to (A), the middle tile to (C) and Wnally the smallest to
(C), a total of seven moves. The solution path thus involves a series of
‘detours’, which form nested subroutines.
We conducted two experiments with this task (Glachan and Light,
1982; Light and Glachan, 1985) using a version very like that shown in
Figure 2.1. In essence, we simply wanted to know whether two
children, performing the task together over a series of trials, would learn
how to do it better than a single child would. As far as the children’s
learning outcomes were concerned, would ‘two heads be better than
one’?
Our Wrst study was conducted with twenty-eight eight-year-olds. All
were given a Wrst session with the task (a ‘pre-test’) on their own, in
which they completed three trials with two-tile towers (which are very
easy) and then three trials with three-tile towers. Their pre-test scores
were based on the three-tile trials, all of which began with the tiles on one
of the outer pegs.
Two weeks later, the children all had a second session, which we can
call the practice session. This time twelve were assigned (at random) to
work alone, while the remaining sixteen children worked in eight pairs,
again with assignment being random except that all were same-sex pairs.
On this occasion they had eight trials on the Tower of Hanoi, all with
three tiles and all beginning on the centre peg, with the ‘goal’ peg
alternating from trial to trial. They were told to try to solve the problem in
the fewest possible moves, and that the minimum possible number of
moves was seven. The children working in pairs sat opposite one another,
at opposite sides of the puzzle. They were told that tile moving must be
done by both of them together, one holding each handle. A week later, all
children had a Wnal post-test session which was individual and identical to
the pre-test.
Analysis of strategies for solving the task suggested a number of more
or less eYcient strategies, taking seven, nine, eleven and thirteen moves.
Solutions taking eight, ten, twelve and fourteen moves reXected use of
one of these strategies combined with a corrected error at some point.
Solutions taking more than fourteen moves all seemed to contain hap-
hazard sequences of moves, and were treated as ‘non-strategic’.
Children were classiWed by their predominant solution patterns at
pre-test.
The study revealed that there was signiWcantly greater improvement
from pre-test to post-test for those children who had been paired for the
practice session than for those who had worked alone. Closer analysis
indicated that this diVerential beneWt from working in pairs applied only
to those children who showed some sort of strategic approach to the
problem at the pre-test stage. Since children were assigned to pairs at
random, most of the pairs comprised children with diVering pre-test
performances. We looked at those children who had been the ‘better’
member of their pair, in terms of pre-test performance, and matched
them up with children from the individual condition with similar pre-test
performance. Those who had worked in pairs still showed signiWcantly
greater pre- to post-test improvement, indicating that in at least some
circumstances the experience of working in pairs can be advantageous
even if one’s partner is less able than oneself.
The second Tower of Hanoi study

A second study replicated the same sequence of sessions as the Wrst, but in
this case all children were paired in the second session. Eighty children
took part, again eight year olds. With all allocations being random, within
sex, some of the children worked under the same instructions as before
(‘structured interaction’), for others the requirement to move each tile
together, using the handles, was not introduced (‘unstructured interac-
tion’). For the remainder, an adult coached the pairs in the correct seven
move solution (‘instruction’).
This time, pre- and post-test trials included both the ‘centre-peg start’
trials used in the second session and the ‘end-peg start’ trials used in the
previous pre- and post-tests. All types of second session were equally
eVective in terms of producing good post-test performances on the
‘centre-peg start’ trials actually used in the second session. However, the
structured interaction condition was signiWcantly superior to the other
conditions as regards pre- to post-test improvement on the ‘end-peg start’
trials. In fact it seemed that only children in the structured interaction
condition were able to extend what they had learned to related problems.
These results lend themselves quite well to interpretation in terms of
the productive potential of cognitive conXict. The fact that children
lacking any strategic approach to the task were relatively unaVected by the
various manipulations is consistent with the idea that one has to have a
viewpoint of one’s own in order to experience a conXict between this
viewpoint and that of a partner. In order to experience conXict, children
must be pursuing some kind of plan or strategy of their own.
The ineVectiveness of direct coaching (which was done by modelling
solutions rather than by explanations) in producing generalised improve-
ment may be seen in terms of the children’s lack of opportunity to
experiment and resolve conXicts between the modelled solution and their
own previous strategy.
The superiority of the ‘structured’ over the ‘unstructured’ interaction
condition is illuminated by the observation (from videotape analysis) that
while both children appeared to be genuinely engaged in all moves in the
structured condition, almost 90 per cent of moves in the unstructured
condition appeared to be determined and carried out entirely by one
child. There was little discussion and almost no indication of joint deci-
sion making. Moreover, many of these pairs were marked by a high
degree of dominance. In the majority of the unstructured pairs, one of the
children determined at least twice as many moves as the other. In this
situation the dominant individual would have experienced little conXict,
while the submissive individual would have little opportunity to resolve
any conXict experienced.
These Wrst two studies, then, attest to the potential value of peer
interaction in problem solving, using a task remote from the Piagetian
repertoire. They indicated that the productivity of this form of interaction
depends on the parties to the interaction having deWnite ideas of their own
about how to approach the task, and having an opportunity to express
these. They make clear that the presentation of a correct solution to the
problem is neither a necessary nor a suYcient condition to produce
progress. Under some circumstances at least, it seemed, two wrongs
could indeed make a right.
The third Tower of Hanoi study

A further study, conducted with a very diVerent version of the Tower of
Hanoi task, allowed us some further insights into the conditions under
which collaborative problem solving with this task could be productive of
good learning outcomes for children (Light, Foot, Colbourn and McClel-
land, 1987). For this study, we employed a computer-based version of the
task, the screen presentation for which is shown in Figure 2.2. The object
and the rules remain the same, but now making a ‘move’ involves keying
in a number to indicate the position from which a disc is to be removed
and then another number to indicate where it should be placed.
One of the attractions of this version of the task is that it allows
automatic collection of move sequence data, and of course it also allows
illegal moves to be prevented automatically. However, unless supple-
mentary constraints are built in, it does not ‘force’ the children’s joint
engagement in each move. In a replication and extension of the previous
study, we attempted to compare the eVectiveness of paired interaction
around this version of the problem with and without such supplementary
constraints.
Sixty children between eight and nine years of age participated in this
study. All were pre-tested individually on the task. After a brief demon-
stration they had two trials, both with three ‘tiles’, moving from position 1
to 2 and then from 2 to 3. On the second session, a week later, the
children were given the same two trials, three times each, each time trying
to complete the task in the smallest possible number of moves. Following
random assignment, twenty children worked individually and twenty
worked in pairs under instructions to work out each move together and to
agree it before moving anything. This condition was referred to as ‘un-
structured interaction’.
The remaining twenty children were assigned to a ‘structured interac-
tion’ condition in which the same general instructions were given, but in
addition the keyboard was constrained so as to accept instructions only
when they were keyed in simultaneously from two separate keypads (one
‘belonging’ to each child), specially labelled up at opposite ends of the
keyboard. Thus the children were eVectively forced to agree the entry,
and both of them had to be involved in the execution of each move.
This reproduces the same type of constraint represented by the require-
ment that each child hold the handles on the tiles in the physical version
of the task.
Individual post-tests included the two trials that the children had
worked on previously, together with two further trials, involving moving
from position 3 to 1, and from 2 to 1. In terms of performance on the
Figure 2.2 The computer version of the Tower of Hanoi
practice session (session two), the pairs did rather better than the individ-
uals, with nothing to choose between the two types of pairing. However,
at post-test, fully correct (seven move) solutions were signiWcantly and
substantially more frequent in the structured condition than in either of
the others, though only for the trials which reproduced the moves prac-
tised in the earlier sessions.
It seems, then, that what the children are eVectively learning in the
structured condition is not a generalised solution strategy, but rather a
particular sequence of moves. Nonetheless, the fact that only the struc-
tured condition led to any detectable superiority over individual practice
conWrms the indications from the previous study that it is not simply
exposure to solutions that diVer from your own, but actually being
forced to engage with them in some way, which fosters improved per-
formance.
Children working collaboratively on the Tower of Hanoi task, whether
using the ‘physical’ or the computer version, did not typically talk to one
another very much. This is of some interest in that it suggests that talk is
not the ‘be all and end all’ of collaborative learning, an issue to which we
shall return in later chapters. But at the time we saw the dearth of
task-related talk as a drawback, inasmuch as it made it diYcult to exam-
ine relationships between the quality of interaction in a particular pair and
the progress made by the members of that pair. Talk is relatively easy to
‘collect’, and more amenable to analysis than non-verbal aspects of
interaction. For this reason, amongst others, we extended our research
programme to a rather diVerent type of problem, namely a code-breaking
problem based on the pegboard game ‘Mastermind’.
The Wrst Mastermind study

A version of the ‘Mastermind’ game was available commercially in the
form of a microchip-based device called ‘Logic 5’, and this was used in
our Wrst study of this problem. The device selected a three-digit sequence
at random, and the task was to determine this sequence by entering
three-digit sequences via a keypad and getting feedback via a visual
display. The feedback informed the player(s) of the number of correct
digits and the number of these that were in the correct position. Pilot
work suggested that this was something children of a range of ages would
be interested in and that, when pairs of children were working together on
it, it would elicit a good deal of task-related talk.
We worked with children of two age levels for this study; a class of seven
year olds and a class of twelve to thirteen year olds, sixty-four children in
all. The initial individual session was more of a training session than a
pre-test. The task was explained to the children, with particular care
being taken to make sure that they understood the meaning of the
feedback.
The children were shown how to record each of their proposed se-
quences on a paper and pencil record pad before entering it. Then, they
had two practice turns on the task, being allowed up to twenty ‘guesses’
to break each code. However, since the generated code sequences
were genuinely random, and some were considerably harder to ‘break’
than others, this introduction did not provide a reliable pre-test score for
each child.
Within each of the age levels, children were then allocated at random to
either an individual condition or a single-sex pair condition for the second
session. In this session they were allowed forty ‘entries’, with which they
had to solve as many problems as possible. Those working in pairs were
told that they had to work together, to agree each entry before recording it
on the sheet, and then take turns in actually keying in the numbers.
Video-recordings were made of the paired sessions. A week later all the
children were given an individual post-test session where they were given
four problems to solve in a maximum of twenty entries each.
The results showed a signiWcant advantage at post-test for those
children who had worked in pairs at stage two. These children required a
mean of 8 entries in the case of the older children and 11 entries in the
case of the younger children to solve each problem. Those who had
worked alone throughout required 11 and 15 entries respectively. Thus
the seven year olds who had worked in pairs were as good at the task
at post-test as the twelve- to thirteen-year-olds who had worked on
their own.
There were also promising indications from the transcribed video-
recordings that the nature and extent of the talk that went on between the
children in the paired sessions was correlated with these children’s post-
test performances. However, the lack of any adequate measure of pre-test
performance in this study meant that it was not possible to establish a
causal relationship; it might simply have been a reXection of children with
a better grasp of the task talking more. An adequate pre-test required us
to be able to specify simpler, introductory two-digit ‘code’ sequences, and
then to control the sequences selected to ensure comparability across
participants. We achieved these aims by creating our own microcom-
puter-based version of the task for a second study.
The second Mastermind study

This study involved seven and eight year olds, forty in all. All were
pre-tested individually using two-digit sequences to introduce the task
and three standard three-digit sequences to get a pre-test score. All
children were then allocated randomly to same-sex pairs for a second
session, which was run just as in the previous study. Finally, all children
had an individual post-test similar to the pre-test (Light and Glachan,
1985).
Videotapes of the second, paired, session were used as a basis for
categorising the pairs into ‘high argumentation’ and ‘low argumentation’
pairs. An argument, for these purposes, referred to an identiWable diVer-
ence of opinion resolved by some form of explanation or justiWcation
related to the problem. Such arguments in many cases included several
verbal ‘counters’, along with explanations, inferences, and so on. As it
transpired, ten of the pairs fell into a high argumentation group, with
more than ten such arguments, while the other ten low argumentation
pairs all had means of fewer than Wve arguments.
The high argumentation and low argumentation pairs turned out not
to diVer in terms of the pre-test scores of the children concerned. Indeed,
the pre-test scores of those who ended up in the low argumentation pairs
were very slightly better. However, the individual post-tests results
showed better performance from the children who had been in high
argumentation pairings, and an analysis of variance showed a signiWcant
interaction between pre- to post-test improvement and the level of argu-

mentation.
Looking more closely at the arguments themselves, it appeared that
what was important was that the children remained focused on the content
of the problem and attempted to resolve their disagreement through
argument. It was relatively unimportant, it seemed, whether the agreed
outcome was the more appropriate of the alternatives under consideration.
The low argumentation pairings were characterised by a tendency to
simply assert a judgement as to what to do next, and if possible to disregard
the partner and press ahead. In the end either one partner took over
completely or turn-taking rules were invoked by the partners.
These results were based upon a very crude typology of verbal argu-
ment, but they do suggest that, with a verbal task such as Mastermind,
aspects of the verbal interaction may give clear pointers to what it is that is
productive about the experience of working in pairs or small groups. This
is an issue which will be explored in a good deal more depth in the
following chapters. Meanwhile there is one more strand of research to be
detailed here, involving a more familiar task, namely the balance beam.
Two balance beam studies

The stages through which children typically progress in achieving an
understanding of how weight and distance inXuence the balancing of a
beam on a fulcrum have long been studied by psychologists (Inhelder and
Piaget, 1958; Siegler, 1976). Since the problem is one in which the
interrelations of several relevant factors need to be taken into account to
achieve a correct understanding, it oVered an attractive candidate for a
study of collaborative problem solving. We conducted several studies
using a computer version of a balance beam task, the basic screen presen-
tation of which is shown in Figure 2.3. On each trial, a beam is presented
with weights already in position, and the task is to predict whether such a
beam would tip, and if so, which way.
The Wrst study using this task was conducted with eighty twelve-year-
old children (Light and Foot, 1987). It involved an individual pre-test for
all children using a paper and pencil version of the task, predicting which
way a series of beams would tip. At stage two the computer version was
used. Half of the children worked individually, while the others worked in
pairs. All of the pairs were asked to work together, but half of them had the
additional requirement that they had to come up with an agreed reason for
their prediction before keying it in. Half of the children working individ-
ually were also asked to come up with a reason for their judgements before
keying them in. The experimenter was on hand as ‘audience’ to the
Figure 2.3 The balance beam task
children’s justiWcations, but these were always received without com-

ment. All children were subsequently individually post-tested.
In terms of pre to post-test improvement, the individual condition
worked least well, and adding the requirement for justiWcations had no
signiWcant beneWcial eVect. The paired condition without justiWcations
was not signiWcantly better than the individual conditions, but the paired
condition with the requirement for justiWcations did produce signiWcantly
better post-test performance than either of the individual conditions. It
turned out on closer inspection that although the children in the ‘pairs
plus justiWcation’ condition were doing better than the others, the strat-
egy which most of them were using involved adding rather than multiply-
ing the eVects of weight and distance. Although this strategy is not
correct, its systematic use in this case resulted in a fairly high proportion
of correct predictions.
Partly as a way of distinguishing strategies, and partly for the purposes
of replication, a second study was conducted using a variant of the
previous task. Here, a beam was presented with weights at given positions
to one side of the fulcrum only. The task was to place weights to the other
side of the fulcrum in such a way as to balance the beam. The direct
mirroring of the weights and positions on the two sides was not allowed,
and the software would not execute it, so the children had to Wnd an
alternative solution.
Once again, eighty children of around twelve years of age took part, and
the conditions were just the same as in the previous study. The results
showed even more clearly than in the previous study that neither pairing
nor justiWcation had much impact on their own, but when combined in
the ‘pairs plus justiWcation’ condition the result was signiWcantly better.
Average pre- to post-test improvement in the ‘pairs plus justiWcation’
condition was in fact more than twice as great as for any of the other
conditions (Light and Foot, 1987).
Friendship and productivity in peer interaction

Despite frequent suggestions that friendship pairs should Wnd it easier to
cooperate and thus beneWt from peer interaction, there is little consistent
evidence on this point. Berndt, Perry and Miller (1988), for example,
found no diVerence in the gains made by friends and non-friends working
together, while Azmitia and Montgomery (1993) did Wnd beneWts for
‘friendship pairs’ as compared with ‘acquaintance pairs’ on at least some
of the scientiWc reasoning tasks studied.
This variable was addressed in one of our own smaller studies (Light,
Foot, Colbourn and McClelland, 1987). Eleven year olds were asked to
nominate the three classmates with whom they most liked to work (note
that this is not quite the same thing as friendship, but undoubtedly
covaries with it strongly). The children were assigned to same sex groups
of three. Four of these groups (referred to as mutually nominating
groups) included only children who had nominated both their partners,
while the other four groups included no partner nominations. The groups
were matched for gender and for general cognitive abilities, using the
results of a school test recently administered to all the children.
The groups of children were given worksheet geography problems
which they had to answer using a computer database. They were encour-
aged to work collaboratively, and were given three twenty-minute
sessions on successive days to work on the task. Video-recordings of
interactions in the groups were scored (at the level of individuals) for such
behaviours as ignoring, agreeing with or disagreeing with a proposal, and
reference to the worksheet. None of these measures indicated a signiW-
cant diVerence between the mutually nominating groups and the others.
The mutually nominating groups sought less help from the experimen-
ter/teacher, and related to this was the Wnding that these groups made
signiWcantly fewer keyboard errors. There was no indication of diVerence
in terms of the children’s grasp of the fundamentals of the task, but the
relatively low level of minor ‘glitches’ in the mutually nominating groups
did seem indicative of more eVective mutual support, or perhaps lower
anxiety. There were no signiWcant diVerences in terms of expressed

attitudes towards the task at the end of the study.
Overview
As will be apparent, the studies described in this chapter involved a fairly
arbitrary choice of tasks. Most were tasks which had been used by other
students of problem solving, and were essentially intellectual rather than
sensorimotor problems. The Tower of Hanoi task demanded forward
planning, while Mastermind required a systematic use of feedback and a
recognition of the signiWcance of negative evidence. The balance beam
comes nearest to the kinds of Piagetian task discussed earlier in this
chapter, in that it demands an ability to recognise and coordinate the
contributions of two discrete variables to the behaviour of the beam.
If the tasks were more heterogeneous than in much previous work, so
too was the age range. The children involved in these studies ranged from
seven to thirteen years of age, but there was little indication that this in
itself was a signiWcant factor in shaping the results. The tight tie between
the mechanism of peer facilitation and a particular phase of intellectual
development, suggested by the socio-cognitive conXict theory, sits rather
uneasily with our results. It seems rather that whatever processes are
involved operate in a rather similar fashion across a considerable range
both of tasks and age levels.
Broadly, the Wndings of these studies do seem to conWrm that the
pairing of children while they work on a problem can improve their
performance on similar problems when they meet them on their own at a
later date. The Wrst Tower of Hanoi study showed that such facilitation
can occur even for the initially more able member of a pair of children,
but is likely to occur only for children who from the outset have at least
some ideas of their own about how to tackle the problem. The second
Tower of Hanoi study indicated that the beneWts of peer interaction
depended on keeping the children jointly engaged and preventing domi-
nance and ‘turn-taking’. The third Tower of Hanoi study conWrmed this
Wnding in relation to a computer-based version of the task, though the
gains here were limited to the speciWc trials used in the paired session.
The Tower of Hanoi studies were characterised by very low levels of
task-related talk between the members even of the most eVective pairs. By
contrast, Mastermind elicited high levels of discussion, and, in some
pairings, high levels of disagreement about what entries to make and why.
The level of such argument turned out to be associated quite clearly with
the pre- to post-test gains made by the members of such pairs. Similarly,
the balance beam task lent itself to verbal justiWcation of the proposed
solutions, and the two studies reported here both suggest that it was the
combination of the availability of a partner and the requirement for such
justiWcation which resulted in the highest levels of progress. The fact that
the verbal articulation of a justiWcation was not in itself suYcient to
produce such progress suggests that it is the process of agreeing a justiW-
cation that holds the key to the productivity of this condition. Reconciling
disagreements might of course be a part of this process.
Thus, the Wndings reported here are consistent with the idea of produc-
tive conXict of views between partners, but the conception of conXict
required is rather broader than Doise’s original socio-cognitive conXict
hypothesis seemed to suggest. The presence of diVering points of view
may be a necessary ingredient of productive conXict, in this sense, but so
too may be a means of securing and sustaining the joint engagement of all
the parties in the resolution of disagreements. The focus on verbal inter-
action and on the co-construction of solutions to tasks, more characteris-
tic of Vygotskian approaches, also Wnds some support in these studies.
Of course, the series of studies revisited here represent only one of
many possible approaches to the empirical study of peer interaction of
children’s learning, and are subject to many limitations. The studies were
conducted in school, but not in the context of ongoing classroom activity,
nor by the regular classroom teacher. For the most part they involved
random allocation of children to experimental conditions, thus riding
roughshod over such considerations as friendship patterns. The Wnal
study suggested that this was not an irrelevant consideration.
Most of the studies described also focused not on pair or group per-
formance as such, but on individual pre- to post-test performance
changes. These were used as a measure of facilitation of individual
learning as a result of typically single and rather brief periods of paired
interaction around a well-deWned task.
We shall trace the further development of this line of research work in
Chapter 4, but meanwhile in the following chapter we shall turn to look at
a speciWc context for learning, namely learning with computers. The
studies described above already reXect a movement towards task presen-
tation using computers, which has a variety of practical advantages for the
researcher. By the late 1980s, it became apparent that computers were
going to Wgure prominently in the learning experiences of both children
and adults, and researchers began to show considerable interest in the
special characteristics and possibilities of this kind of learning. This
interest has shaped the direction of subsequent research on social pro-
cesses in learning to a considerable degree, so a brief review of the
‘computers and learning’ literature seems appropriate at this juncture.
3 Computers and learning
Introduction
A comprehensive review of the research literature on what computers
might have to oVer for learning would have to go back a long way to Wnd
its starting points (e.g. Suppes, 1966; see also Light, 1997). However, at
least in the in the UK, it is only relatively recently that children have been
able to secure reliable access to at least one classroom computer, most
teachers have received some training in the use of computers, and the
school curriculum envisages computer use in most subject areas (Crook,
1994). The prominence given to computers, and the substantial invest-
ment of resources entailed in their use, are widespread features of educa-
tion across the developed world. The psychological theories of learning
which have informed the development of educational computer use over
the last thirty years or so oVer a fair reXection of the psychology of
learning more generally across this period. In this chapter we shall explore
some of the ways in which psychological accounts of children’s learning
have contributed to, and been reXected in, the way computer-based
learning has developed.
Computers and instruction

Much of the software used in schools, often referred to under the heading
of computer-assisted instruction (CAI), owes its origins at least in part to
the associationist learning theory tradition in psychology. A leading expo-
nent of that tradition, B.F. Skinner (e.g. 1965), was heavily committed to
the idea that machines could be developed which would teach children
more eVectively than the classroom teacher was able to do. The ‘teaching
machines’ of the 1950s and 1960s were largely unsuccessful, but the
development of the microcomputer oVered a new lease of life to this idea.
A traditional learning-theory approach involves focusing very narrowly
upon achieving some desired pattern of overt behaviour. The generation
of desired behaviour patterns is seen in terms of progressive shaping
27
through small incremental steps. The principal mechanism for achieving

such shaping is envisaged as being the reinforcement of correct responses
through the reliable delivery of rewards.
In contemporary CAI, the idea is that the computer assists the teacher
to achieve a speciWc instructional goal. The classroom teacher cannot give
the level of attention to the individual learner that would be required to
implement an individualised instructional programme, but the computer
can. With CAI, pupils can receive instruction tailored to their individual
level, and their own pace of learning. Much of the relevant software oVers
‘drill and practice’ on well-deWned skills in areas such as arithmetic and
spelling. The ‘reinforcement’ for success usually relates to some game-
like activity which runs in parallel to the routine tasks, but depends
on success in them. The tasks typically consist of numerous Wxed for-
mat items graded in diYculty, with progression depending on correct
responses.
This general approach dominated early use of computers in schools,
and software in this tradition is probably still more widely used than
anything else, at least for children under about nine years of age (Crook,
1994). It is not easy to evaluate the eVectiveness of such software, since
comparative studies necessarily involve comparison of particular pieces of
software with particular alternative ways of teaching the same material.
Nonetheless, the available reviews and meta-analyses suggest moderate
but worthwhile gains for CAI-taught pupils as compared with ‘chalk-and-
talk’ alternatives (e.g. Niemiec and Walberg, 1987).
Whereas Skinner saw reinforcement of correct responses as the key to
progress, contemporary CAI software usually oVers some corrective
guidance, which depends on the nature of the error made. With most CAI
software, however, the level of ‘error diagnosis’ undertaken by the com-
puter is minimal. More sophisticated versions of such tutorial software
attempt to take account of the pattern of errors made by the child, thus
modelling the child’s reasoning. A model of the learner’s current under-
standing can then in principle be used to select the most appropriate
further tasks or instructions. Such an ‘intelligent tutoring system’ is thus
eVectively learning about the learner in order to teach.
Intelligent tutoring systems developed to date, however, typically rest
on the assumption that the learner’s knowledge consists of a subset of
adult knowledge, or that the learner is using one of a pre-determined list
of erroneous rules (Costa, 1991). Even in the limited areas in which
adequate psychological models of children’s learning are available, it is
not as yet established that in practice an intelligent tutoring system is
more eVective than an unintelligent tutoring program based on the same
learning model (Nathan and Resnick, 1994).
Computers and learning 29
Computers and construction

Just as in developmental and educational psychology more generally,
advocates of associationist conceptions of learning with computers found
themselves confronted with radically diVerent conceptions of develop-
ment and learning emanating from Piagetian theory. From a construc-
tivist standpoint, of course, the key to education lies not in shaping
behaviours, but in providing the conditions in which children themselves
can construct an understanding of their world. The teacher is seen, not as
instructor, but as facilitator, providing a stimulating and encouraging
environment for intellectual exploration. In just the same way, within this
tradition the role of the computer is not to teach but rather to aVord
opportunities for constructive activity.
Papert (1980, 1994) in particular has argued that the potential of the
computer lies in extending children’s control over their own learning.
The computer should not be programming the child, he argued, rather
the child should be programming the computer. To this end, Papert
developed a modular programming language called Logo, suitable for use
by even quite young children. Programming in Logo is seen as linking the
child’s intuitive grasp of how to do things in the world with the develop-
ment of abstract levels of understanding of the same phenomena.
Working with the computer in this way is seen as having the potential to
foster ‘powerful ideas’, in the form of generalisable thinking and problem-
solving skills. We saw in Chapter 1 that Piaget held the view that children
cannot develop their thinking as eVectively in interactions with adults as
they can with their peers because of the confounding eVects of adult
authority. The same stance is reXected in Papert’s assertion that
children’s Logo programmes should not be judged right or wrong accord-
ing to the canons of adult authority. Rather, the program either works or
it doesn’t. If it doesn’t, a process of ‘de-bugging’ is required, which
involves critically re-appraising the way it has been constructed.
Logo has established a place for itself in schools, but mainly as a useful
approach to learning some aspects of mathematics (e.g. Hoyles and
Sutherland, 1989). As far as fostering generalisable problem-solving skills
is concerned, evidence of success is limited to situations in which Logo is
used in the context of an intensive and highly structured curriculum, put
together with this aim in mind (Pea and Kurland, 1984; De Corte,
VerschaVel and Schrooten, 1992).
The constructivist approach to learning is evident in a number of other
approaches to computer-based learning. An emphasis on ‘learning by
doing’, and on using the computer as a vehicle for spontaneous explora-
tory learning is apparent in the recent enthusiasm for adventure games
and simulations. Schank and Cleary (1995), for example, draw an anal-
ogy with Xight simulators, which assist pilots to learn to Xy by providing
realistic environments without all the risks and costs of the real thing. In
much the same way, they argue, other types of experience can be
simulated on the computer, allowing children to play, experiment and
explore, and greatly extending the range of things children can learn
by doing.
Multimedia computer simulations can allow children to explore the
inside of the human body, or the outer reaches of the universe. ‘Alterna-
tive’ worlds can be simulated, in which, for example, diVerent laws of
motion apply (Smith, 1991). Developments in the technology of virtual
reality may extend the possibilities of such ‘simulated worlds’ very rapidly
in the future. Equally, other developments in the Weld of hypermedia are
already bringing to the classroom a much wider range of learning re-
sources, along with an educational philosophy which stresses the need to
‘put the learner in control’ (Hutchins, Hall and Colbourn, 1993).
Computers and collaboration

As we saw in Chapter 1, in developmental psychology Piagetian construc-
tivism has been somewhat overtaken by a social constructivist agenda
identiWed with Vygotsky’s ideas. Similarly, in the Weld of computers and
learning the social dimensions of computer use are increasingly coming to
centre stage.
Whereas early advocates of computer use in schools emphasised the
beneWts of individualising the learning process, in practice, things have
turned out rather diVerently. In the 1980s classroom observational stu-
dies and surveys began to suggest that most educational use of computers
in practice involves pairs or small groups, rather than individuals, and
that these pairs or small groups often work relatively independently of the
class teacher (e.g. Cummings, 1985; Jackson, Fletcher and Messer,
1986). Some classroom-based studies have claimed that small groups
focused around a computer task are able to sustain task-related interac-
tion better than similar groups working on non-computer tasks (e.g.
Scanlon, IssroV and Murphy, 1999).
Experimental evaluations of collaborative as against individual modes
of learning with computers stem from a number of diVerent research
traditions, and relate to a number of diVerent types of computer-based
learning experience. Already in the pre-computer ‘teaching machine’
literature, there was evidence that, despite the supposed importance of
self-pacing, students learned as well or better in small groups as when
working on their own (Hartley, 1968; Hartley and Hogarth, 1971).
Computers and learning 31
Likewise in the context of CAI, despite the importance often attached to

students working at their own individual level and pace, there is evidence
that average individual learning outcomes are better for children who
work together in pairs than for children who work alone (Mevarech,
Silber and Fine, 1991).
If we turn to programming, we saw in our earlier discussion that the key
claim made for Logo was that it supported the development by the
individual of generalisable thinking skills, but that evidence in support of
this claim is scant (Clements, 1987; Kliman, 1985). However, a number
of studies with Logo and other programming languages have found
evidence that peer interaction can facilitate children’s achievements in
programming. Children programming in Logo collaboratively have been
found to do better than individuals on a variety of criteria (Healy, Pozzi
and Hoyles, 1995; Clements and Nastasi, 1992), and the quality of social
interaction and discussion has been shown to predict achievement in
terms of eventual individual programming skills in Basic (Webb, Ender
and Lewis, 1986).
A shift from a ‘Piagetian’ to a ‘Vygotskian’ perspective in considering
computers and learning is further reXected in a recognition, supported by
qualitative studies of computer use in authentic learning environments
(Crook, 1994; Mercer, 1995, SchoWeld, 1995), that collaborative modes
of learning involve teachers as well as peers, and that they take place in
social and institutional frameworks. For the moment, though, our own
focus will remain on the experimental analysis of peer interaction and
learning in the context of children’s computer use. We hope to show,
through the studies described in the following chapters, how such studies
can be used to expose and explore something of the complexity of the
social bases of children’s learning.
4 Kings, Crowns and Honeybears: a second
series of studies
Introduction
In this chapter we shall describe a series of experimental studies based on
the collaborative use of computer-based planning tasks. The studies are
linked by the particular type of task used. Rather than start from existing
tasks, as in the Wrst series of studies (Chapter 2), we decided to develop
our own. Instead of relying on computer simulations of non-computer
tasks, such as the Tower of Hanoi or the balance beam, we wanted to
develop a task which would better exploit the potential of the computer.
We also wanted a task which would allow a more extended experience of
solo or collaborative problem solving, rather than a sequence of quick
trials.
We were interested by evidence from RogoV and colleagues (Gauvain
and RogoV, 1989; Radziszewska and RogoV, 1988) that the ‘metacogni-
tive’ processes involved in successful planning (e.g. conscious, explicit
monitoring and careful control of plan development) might lend them-
selves particularly well to facilitation through interaction, discussion and
collaboration. We were also aware of evidence that ‘adventure games’
seemed to oVer particularly stimulating contexts for peer discussion and
interaction (Crook, 1987; Johnson and Johnson, 1986).
The task we designed lies somewhere along a continuum running from
genuinely open adventure games to closed logical problems of the Tower
of Hanoi type that we had used previously. We wanted a well-structured
problem, but one which at the same time oVered a variety of routes to
solution, some more eYcient than others. We were also seeking to cap-
ture several important aspects common to many situations of learning
with computers. For example, computers may often hold a large amount
of information which needs to be located and used systematically as a
basis for completing the given task. The information is used as a basis for
formulating a plan, which may then need to be amended in the light of
feedback in the course of doing the task. We felt that this combination of
32
Kings, Crowns and Honeybears 33
information searching and planning was likely to Wgure prominently

amongst the skills needed to work and learn with computers.
The Wrst King and Crown study

The King and Crown task is a route-planning task implemented in
Hypercard on a Macintosh computer. Whilst the underlying problem is
fairly simple, the task as a whole is quite complex, as successful comple-
tion requires the user to search for relevant information, plan a solution
and react constructively to any obstacles encountered along the way. The
scenario is a quest involving the retrieval of a crown from an island. The
users are told that: ‘The King lives in his castle in Ashlan. He wants his
crown and all his subjects (the driver, the guard, the pilot and the captain)
in Ashlan for a feast. He wants you to give the orders to get them all there.’
The initial screen shows a map on which there are a number of
‘buttons’ marked by rectangles (see Figure 4.1). When the mouse-driven
cursor is positioned over and then clicked on one of these buttons, further
screens are presented. Thus if the user clicks on a place name, a screen
appears which allows the user (via further buttons) to access information
concerning the objects, persons and means of transport present at that
particular location.
Users can also gather information by using the ‘Info’ button, which
accesses a general information screen (see Figure 4.2). This oVers an
alternative way of obtaining information. If, for example, the users click
on the button marked ‘Pilot’ on the ‘Info’ screen, they will obtain another
screen allowing them to obtain information about what the pilot can do
and/or to discover his whereabouts (see Figure 4.3). The highly embed-
ded nature of these information resources means that users are given only
the information they speciWcally ask for, and also that we can keep track of
exactly what information they have obtained, and when.
The ‘Goal’ button makes available a written statement of the goal or
aim of the game, whilst the ‘Key’ button makes available information on
the various route markings. Thus, through an extended process of infor-
mation searching, the users can discover the initial whereabouts of the
characters, the crown and the diVerent means of transport. In fact in the
standard version of the task the driver, guard, pilot and captain all turn
out to be initially at Ashlan. The crown is on the island of Fruggle,
however, and so needs to be retrieved. There is a car at Ashlan, a ship at
Brockley, another ship at Crowmarket and a plane on the island of
Hushley.
When the users think they have suYcient information, they can initiate
a move by clicking on the ‘Act’ button. This accesses the screen shown in
Figure 4.1 King and Crown software: the map
Figure 4.2 King and Crown software: the general information screen
Figure 4.3 King and Crown software: the pilot information screen
Figure 4.4 King and Crown software: the action screen

Figure 4.4. They are then required to specify the point of departure, a
destination, the characters being moved (and whether or not they are
moving the crown) and the means of transport being used. The move is
then made by clicking on ‘Go’.
The task is made complicated by the presence of bandits and pirates.
The bandits will steal the crown on the mainland unless the guard is
present. The pirates will steal the crown from any ship sailing the sea even
if the guard is present. If the users obtained this information at the outset
and thought it through, they could correctly conclude that the crown
must be retrieved by plane. However, the plane can only carry a pilot and
one passenger, which complicates things further. The optimal solution, in
fact, is to take the car (and driver, guard, captain and pilot) to Crowmar-
ket; to take the ship (and captain and pilot) to Hushley; to take the plane
(and the pilot and the captain) to Fruggle to collect the crown; to return
to Crowmarket, and then for all to return by car to Ashlan. This repre-
sents a total of Wve moves.
The software (initially developed by Richard Joiner) automatically
updates all relevant information as each move is made. Thus at any
point the children can stop and take stock of where the characters and
means of transport are. Whilst in theory this might not be necessary, in
practice it usually is. Most users, at least initially, either set oV in the
wrong direction or take the wrong characters, and thus encounter diY-
culties. If the users attempt to make a move which is impossible, or
which would lead to the pirates stealing the crown, they get a warning
message and are prevented from actually making the move. This means
that the users then have to replan the move, to take account of the
particular problem they have encountered. The task is thus a diYcult
one. Nevertheless most users, whether children or adults, Wnd it engag-
ing and highly motivating.
In an initial study using the ‘King and Crown’ software we set out to
investigate whether children of around eleven years of age would proWt
from working collaboratively on this task (Blaye, Light, Joiner and Shel-
don, 1991). First we wanted to know whether, on a task such as this
which makes heavy demands on information management and planning
skills, those working in pairs would perform better than those working
alone. Second, we wanted to see to what extent any advantage shown
by the pairs would transfer to a subsequent assessment of individual
performance.
The study involved thirty-nine eleven-year-old children. They were
taken from their school classroom to another room in the school either
in pairs or individually (thirteen pairs and thirteen individuals). The
decision as to which should work alone and which in pairs was made
randomly. Allocation to particular pairs was also random except that

pairs were always same-gender, and the teacher’s advice was taken so as
to avoid pairing children who actively disliked one another.
The children sat at the computer, which was controlled by mouse only.
A video camera stood nearby on a tripod, and all paired sessions were
videotaped. Children were given a scripted introduction to the object of
the task, and (if necessary) to using a mouse. They were given a brief
demonstration of how to obtain information and how to make a move,
and completed a short practice task. They were provided with a paper
copy of the map for reference, and encouraged to share the mouse and to
work together to solve the problem.
The children were given thirty minutes to work on the task, during
which time the researcher was present but very much in the background.
They were told in advance not to worry if they didn’t Wnish it, as they
would have another go the next week. A week later they did have a second
session, in pairs or individually just as in the Wrst session. The task was the
same, starting at the beginning again, and this time they had twenty-Wve
minutes (unless they succeeded before this). They were told to try to
complete the task in as few moves as possible.
A week later again, all children had a third session, but this time all of
them worked individually. The task was also slightly diVerent, in that the
characters and means of transport were not all in the same places at the
outset. The children were told about this before they started. This was
done to ensure that the children could not succeed just by following a
formula remembered from a previous session. They were given twenty
minutes maximum on this post-test session.
On the Wrst session, none of the thirteen children who worked alone
succeeded in achieving the goal within the time limit. Two of the thirteen
pairs (15 per cent) succeeded. On the second session, 15 per cent of the
individuals and 46 per cent of the pairs succeeded, a diVerence which
reached statistical signiWcance. On the third session all children worked
alone. Of those who had previously worked in pairs, 72 per cent suc-
ceeded, whereas of those who had worked alone throughout, only 31 per
cent succeeded. Again, this diVerence was statistically reliable.
Examination of the computer logs made it clear that most children
made initial moves without any clear idea where they were leading. They
usually had to retrace them when it became apparent that they had not
got the right personnel or means of transport available. By the second
session there was evidence of more planning, signiWcantly more so in the
case of the children working in pairs. In fact all thirteen pairs got as far as
moving the crown on the second session, but only three of the thirteen
individuals did so.
Close scrutiny of the strategies adopted suggests that very few of the
children even in this second session actually planned out a successful
strategy in advance. In fact ten of the pairs and eleven of the individuals
tried to move the crown by ship, but were stopped in their tracks by the
warning message about the pirates. The diVerence between pairs and
individuals was that, following this setback, six of the pairs but only two of
the individuals immediately turned their attention to the plane. Thus it
seems that the advantage of the pairs may lie more in adaptive re-planning
than in thorough pre-planning.
The videotapes were analysed in terms of who, within the pairs, con-
trolled the mouse most, and how decisions were made. Mouse control in
the paired sessions (sessions one and two) turned out not to be correlated
with score on the individual post-test (session three), so it did not seem
that this aspect of the interaction was strongly related to ability or learn-
ing. However, the lack of an individual pre-test meant we were unable to
rule out the possibility that improvement was related to the extent of
mouse control.
Decisions regarding the destination for the next move, or who was to be
taken, or what means of transport was to be used, were examined for all
pairs in the second session. Intercoder reliability was checked on a ran-
dom sample of six pairs. A particular child was identiWed as the author of
a particular decision if he or she acted without reference to the partner or
suggested a choice which the partner acted on. Intercoder reliability was a
little over 80 per cent. Each of the children involved was thus given a
‘decision score’ based on the proportion of all decisions that child was
responsible for. Decision score did turn out to be signiWcantly correlated
(−0.45) with individual performance on the third session. The children
who controlled more of the decisions in the second session tended to do
better as individuals on the third. Decision score was not related to any
tendency to control the mouse more when working with a partner.
Decisions which were formulated simultaneously, or where both part-
ners contributed some part of the formulation, were coded as shared
decisions. It is interesting that shared decisions were signiWcantly more
prevalent in pairs in which both partners subsequently succeeded on their
own. Pairs can also be scored in terms of the extent of disparity in the
decision scores of their members. If we look back to the crucial point in
the task alluded to earlier, where pairs were confronted with the impossi-
bility of taking the crown by ship, it turns out that the six pairs who
‘adaptively replanned’ to take the plane at that point were the six with the
lowest decision disparities. This suggests that it is not just having a
partner that makes the diVerence, but having a fairly symmetrical rela-
tionship with that partner as far as decision making is concerned.
The videotapes also allowed us to see something of the patterning of

role diVerentiation across time on the task. The mouse could only be held
by one child at a time, but the paper map tended to be kept by the child
who was not holding the mouse at that time. By channelling certain kinds
of executive control into one role, the mouse may be seen as opening up a
complementary role of ‘navigator’ or ‘strategist’. A ready analogy might
be two people driving around a strange town looking for an address; one
could be attending to the minutiae of driving the car, while the other
works at how to reach the destination.
Some researchers have commented on the potential disadvantages of
role diVerentiations of this kind, which may lead to a child mastering only
part of what is required to complete a task (e.g. Sheingold, Hawkins and
Char, 1984; Hoyles and Sutherland, 1989). This did not appear to be a
problem here, perhaps because having been encouraged to share the
mouse, the children exchanged roles fairly frequently. Both Sheingold et
al. and Hoyles and Sutherland were describing much longer sequences of
interaction in classroom settings. It may be that over longer periods and
with less supervision, the children tend to settle into less-than-productive
patterns of role division. Interface and task characteristics may also be
important in shaping such patterns.
It should also be acknowledged that other, less cognitive factors may
also be involved. The success of ex-pair members on the third session
could reXect increased self-conWdence based on the prior experience of
paired success. More generally, the presence of a partner on the Wrst two
sessions may have signiWcant aVective and/or motivational eVects which
may in turn have consequences for performance. We shall return to these
issues later.
To summarise the results of this study, then, we can say that (i) the
children working in pairs were at least twice as likely to succeed on the
King and Crown task as the individuals, and (ii) the children who had
previously worked in pairs on the task were at least twice as likely to
succeed on the post-task variant of the King and Crown task as those who
had worked alone throughout. Since the task was novel for all children at
the beginning, success cannot rest on what they knew at the outset. We
are looking at how an understanding of the task domain, the ways of
acting within that domain, and the solution strategy have been built up
more or less successfully over some seventy-Wve minutes of computer-
based work. On the basis of this study it seems that peer interaction can
play a signiWcant part in the construction of this kind of understanding,
even without the kind of ‘forced interaction’ (dual key controls, for
example, see Chapter 2) which we had to resort to in the Wrst series of
studies.
The second King and Crown study

A second study was conducted to allow us to obtain a substantially larger
sample of paired interactions for analysis. Our aim was to identify import-
ant aspects of the interaction, paying particular attention to the knowl-
edge created and shared in the verbal interaction between the partners.
We wanted to see whether we could relate measures of such verbal
interaction to the outcomes for the pairs, and to the pair members’
subsequent performance when working on their own.
Since our interest was in the kinds of interactions that were associated
with more and less successful pairs, we did not use an individual compari-
son group in this study (Barbieri and Light, 1992). Also there were only
two sessions. All of the children (eleven-year-olds again) worked in pairs
in the Wrst session. There were thirty-three pairs in all, eleven pairs of
boys, eleven pairs of girls, and eleven mixed gender pairs. In the Wrst
session the children worked in pairs for twenty-Wve minutes. This session
was video-recorded. In the second session all children worked individ-
ually with a variant of the task, as in the previous study.
The King and Crown software used was amended in several ways for
this study. The bandits were eliminated, as was the guard. The automatic
logging was reWned to provide various summary measures, including a
new measure of levels of success. These levels correspond to signiWcant
points of progression through the task, all being necessary to eventual
success. Thus performance was categorised according to whether the
children got no further than moving the car (level one), moving the ship
(level two), moving the plane (level three), moving the crown (level four),
getting the crown to the mainland (level Wve), or successfully completing
the task (level six). The videotape transcripts, combined with the com-
puter logs, allowed a fairly full picture to be built up of the children’s
interactions with one another and with the computer. The computer
provides both information and feedback, and the ways in which these are
integrated into the children’s interactions was one of our chief concerns.
The categories developed for analysis are given in detail in Barbieri and
Light (1992). Planning was operationalised as verbal reference to a
sequence of moves or possible moves. For example: ‘We’ve got to go to
Ashlan to pick up the captain, and then we’ll take the ship to Fruggle’.
Negotiation was coded when a proposal from one child was opposed by
the partner, and one or both parties went on to defend their point of view
or attempt to overcome the contradiction. For example: ‘Now, back to
Brockley and Ashlan’; ‘No, you can’t’; ‘Yes we can’; ‘No, ‘cos the pirates
will get us’. Integration of information search was coded when children
drew explicit conclusions from the information they obtained from the
computer, integrating it into a plan of action. Similarly, integration of

error messages was coded when explicit statements of implication were
drawn on the basis of error (or warning) messages received from the
computer. All these codes were assigned to pairs (i.e. to segments of
conversation between the two children), rather than to individuals.
The results revealed that planning was the most successful of the
interaction measures in terms of its relationship with levels of perform-
ance, correlating +0.6 with the level of success of the pair on session one,
and +0.4 with the level of individual success (averaged across the two
partners) on session two. Integration of information search showed lower,
but still signiWcant, correlations, at +0.4 and +0.3 respectively. Negoti-
ation correlated with paired (+0.3) but not individual performance, while
integration of error messages correlated only with individual perform-
ance (+0.3).
Another way to look at the data is in terms of the relationships between
the interaction measures and the ‘pairwise patterns’ of success in session
two. The ex-members of any given pair might both succeed (SS), they
might both fail (FF), or one may fail whilst the other succeeds (FS). For
this purpose, performance levels one to three were treated as failure whilst
levels four to six were considered as success.
When we look at the prevalence of the diVerent forms of verbal interac-
tion across these three pair-outcome types, an interesting pattern
emerges. The SS pairs show signiWcantly higher levels of planning, nego-
tiation and integration of information search than either the FS or the FF
pairs. There is no diVerence on any of the interaction measures between
the FS and FF pairs. By contrast, if we look at the levels of success
achieved by the pairs at session one, it is the FS and SS pairs which look
similar, scoring signiWcantly better than the FF pairs. It seems, then, that
the FS pairs are characterised in session one by a low level of constructive
verbal interaction relative to their level of achievement. The achievement,
one assumes, rests upon the performance of the more capable child, but
they are not succeeding in building a shared understanding of the task.
It seems, then, that the verbal measures designed to index planning,
negotiation and the coconstruction of knowledge do relate signiWcantly to
successful problem solving by the pairs, and to learning outcomes as
indexed by the subsequent success of individual children in the second
session. However, to be fully conWdent that these associations reXect
causal processes we would really need access to appropriate pre-test
measures of a kind which were not available in this study. We shall return
to other Wndings from this study when discussing gender issues in the next
chapter. Here we shall move on to a third and Wnal piece of research
conducted with the King and Crown task, this time with adults.
The third King and Crown study

This small scale study was conducted by Agnes Blaye with students at a
French university, using a French language version of the software (Blaye
and Light, 1995). The students had two sessions with the task, working
either in pairs throughout (Wve pairs) or individually throughout (six
individuals). On the Wrst session only two of the individuals got as far as
moving the ship, and none moved the plane, whereas all the pairs moved
the ship and one pair got as far as moving the plane. On the second
session only one individual succeeded in retrieving the crown, whereas all
of the pairs did.
Although small, this study is of interest for two reasons. First, it is
notable that the task itself seems to work just as well for undergraduates as
for eleven year olds, and indeed causes them nearly as much diYculty. Of
course there may be cohort eVects; when the eleven-year-olds we tested in
the Wrst two studies reach university they may have gone beyond this kind
of thing! Secondly, it is interesting to see that the peer advantage which
marked the results of the Wrst King and Crown study is apparent here too,
with university students, although we have no individual post-test
measure in this case.
Even at session two, more than one third of the moves attempted by
the students working alone failed, either because they had not speciWed
all the parameters or because they were attempting something that was
impossible. No such diYculties occurred amongst the pairs. The stu-
dents working alone also made a considerable number of redundant
information searches, that is, searches for information that they had
already in fact obtained. This accounted for 31 per cent of the informa-
tion searched by the individuals on the Wrst session and 25 per cent of
that searched by individuals on the second session. The corresponding
Wgures for redundant information searches by pairs were signiWcantly
lower, at 8 per cent and 0 per cent. This points to the conclusion that the
pairs were better at remembering what they had done and what they had
found out; another important aspect of building a shared representation
of the task.
The various analyses conducted across the three King and Crown
studies have oVered some leads concerning the factors inXuencing suc-
cessful learning, but the signiWcance of many of these cannot really be
assessed without having some idea of the initial competencies of the
individuals, prior to the experience of working together. To address these
issues we conducted a much larger study, building in pre- as well as
post-tests. At the same time, we made changes in procedures. Most
importantly, we changed the nature of the ‘individual’ conditions, as will
be described below. We also developed a new version of the software,

which we called ‘Honeybears’.
The Wrst Honeybears study

The Honeybears task was designed to eliminate some of the more obvi-
ously male-stereotyped elements of the King and Crown task, for reasons
discussed in more detail in the next chapter. The underlying structure of
the Honeybears task is, however, exactly the same as the King and Crown
task. In the Honeybears version, the scenario is that three bears (referred
to as the ‘honeybears’) have set out for a picnic at Almwood, and discover
that they have forgotten their honey. The honey is on an island at the
other side of the river, but in the river there are honeymonsters who will
steal the honey if the bears attempt to retrieve it by boat (honeymonsters
Wgured at the time on advertisements for a popular honey-coated break-
fast cereal, so the children could conjure up an image of them quite
easily). The presence of the honeymonsters necessitated the retrieval of
the honey by air. In this scenario we used a pony rather than a car, a
rowing boat rather than a ship, and a hot air balloon rather than a plane.
Each of the bears was able to handle one of these means of transport, so
we had ‘ponybear’, ‘waterbear’ and ‘airbear’.
The basic map screen is shown in Figure 4.5. The optimal solution
involves (i) taking all three bears by pony from Almwood to Cob, (ii)
taking ‘airbear’ and ‘waterbear’ to Hilt by boat, (iii) taking both on to
Flint by balloon, (iv) returning by balloon to Cob with the honey, and (v)
bringing all three bears and the honey back to Almwood by pony. Thus,
whilst the Honeybears and the King and Crown tasks diVer with respect
to the context in which the problem is set (and also with respect to certain
minor interface characteristics), both present the same problem in the
same adventure game format and call for an identical solution strategy.
For the Wrst study using the ‘Honeybears’ variant of the software
(Littleton, Light, Joiner, Messer and Barnes, 1992), we incorporated an
individual pre-test session to provide us with an index of each child’s
initial level of competence. This makes it possible to take account of
antecedent diVerences in children’s competence when examining the
association between task performance and other variables. The incorpor-
ation of such a session also allows us to assess the extent to which
diVerences in outcome might reXect antecedent diVerences in the relative
abilities of the two members making up a given pair. The study also
incorporated an ‘individual’ condition, allowing us to compare the per-
formance of those who had worked in pairs at some stage with those who
worked alone throughout.
Figure 4.5 Honeybears software: the map
From observation in the context of other studies, we had reason to

suppose that some of the children who worked at the computer alone in
the Wrst session were apprehensive, and we wanted to make the situation
more ‘user friendly’. Since we by now had the resources to set up four
computers in the schools instead of one, we took children from the
classroom in fours. Where they were to work alone, each sat at a computer
and worked without conferring with the others. Where they were to work
in pairs, members of each pair sat together at a computer, and the other
two machines went unused. The positioning of the computers around the
periphery of the classroom was such that the screen of each was visible
only to its user(s). Nevertheless, the children were always working in the
presence of other children who were similarly engaged.
One hundred and twenty eleven and twelve-year-olds took part in the
study. The Wrst stage involved each of them having a Wfteen minute
session working on their own with Honeybears. From this we were able to
derive an index of each child’s initial level of ability on the task. The
second stage, which took place a week later, consisted of a thirty-minute
session in which the children once again worked on the Honeybears task.
Thirty of the children continued to work alone and the others worked
with a partner.
Children’s performance at the Wrst stage was not used in assigning
children to conditions for the second. Children were allocated at random
to the paired or individual conditions, and then within the group of

children allocated to paired conditions, their teachers were invited to
suggest pairings of children who could be expected to work well together.
Children working as a pair always came from the same school class. There
were forty-Wve pairs in all, with a mix of within-gender and between-
gender pairings (this aspect of the study will be picked up in Chapter 5).
At the third stage of the study, a week later, all 120 children were
post-tested individually in a twenty-minute session using a variant of the
Honeybears task in which initial locations were changed.
The children’s performance on the task was measured using a revised
version of the scoring system employed in the second King and Crown
study. This new scoring system introduced a level 0, which indicates that
the children had failed to make a move at all. Levels one to six remained
unchanged, whilst levels seven and eight were added to give extra credit to
children who completed the task either more quickly or in a smaller
number of moves.
Figure 4.6 shows the overall results, comparing those who worked
alone throughout with those who were paired for the second stage.
Predictably, given the random allocation to conditions, there was no
diVerence in performance between the groups at the Wrst stage. Figure 4.6
suggests that there was an advantage for pairs over individuals at the
second stage, but that there was no carry-over of this advantage to
individual post-test performance.
Since we now have individual pre-test performance measures, we can
test the signiWcance of the pair advantage by looking at the relative
improvement of those who worked in pairs as against those who worked
alone. An analysis based on change of score from Wrst to second stage
showed a signiWcant advantage for those who worked in pairs. An alterna-
tive statistical procedure using the mean change score for each pair
yielded a similar result.
Mean performance at the post-test stage was somewhat poorer than at
the second stage. However, it needs to be remembered that the post-test
used a variant of the task, and also allowed less time. The important
observation is that, in contrast with the Wndings of the Wrst King and
Crown study, we have no evidence here of a beneWt at post-test for those
children who had earlier worked in pairs.
As noted earlier, the allocations of children to particular pairs at stage
two were made without regard to the children’s pre-test (stage one)
scores. Consequently, some children were assigned to a partner who had
scored the same as they had themselves at pre-test. These may be termed
‘symmetrical’ pairs. Other children were assigned to pairs in which there
was a diVerence of one performance level in the pre-test (‘asymmetrical’
Figure 4.6 The mean level of performance for the ‘pairs’ and ‘individ-
uals’ during each of the three sessions
pairs). In others there was a diVerence of more than one level in the
pre-test performance of the pair members (‘highly asymmetrical’ pairs).
As it turned out, the pairs established at stage two actually comprised
fourteen symmetrical pairs, twenty-three asymmetrical pairs and eight
highly asymmetrical pairs. We analysed the progress made (from stage
one to stage three) for members of each of these types of pair. The average
improvement in performance level from stage one to stage three averaged
3.8 for members of symmetrical pairs, 2.9 for members of asymmetrical
pairs and 2.4 for members of highly asymmetrical pairs. This diVerence
proved to be statistically signiWcant overall. The outcome for the symmet-
rical pairs was signiWcantly better than for either of the other two groups.
As a check on this rather striking result, we went on to examine the
performance of a sub-sample of children who all started from the same
pre-test level. The modal pre-test score (level one) was used for this
purpose. Thus, one group consisted of those children whose pre-test
score was one and who were paired with a child whose pre-test score was
less than their own (n = 7). Another group consisted of those whose
pre-test score was one and whose partners’ score was greater than their
own (n = 16). The third group were those whose partner also scored one
at pre-test (n = 12).
Figure 4.7 shows the results of a comparison between these three
groups. The best stage two performance came from those children work-
Figure 4.7 The mean session 1, 2 and 3 performance of the ‘same’,

‘less’ and ‘more’ able pairings
ing as symmetrical pairs. By stage three the performance levels were

lowest for those who had been paired with a more able partner, not
signiWcantly better for those who had been paired with a less able partner,
but signiWcantly better for those with a partner of the same initial ability.
This analysis indicates that ‘symmetrical pairings’ (where the children
are of the same initial level of ability) function better than ‘asymmetrical
pairings’ (where the children are of diVerent initial levels of ability), both
in terms of pair performance and in terms of individual learning out-
comes. With a task like this where the participants are all ‘starting from
scratch’ and having to grapple with complex information (and an unfam-
iliar interface), it seems that a balance of initial abilities may hold the best
promise of a good learning outcome.
Some light is thrown on the processes underpinning this ‘symmetry
advantage’ by an analysis undertaken using an approach to interaction
analysis based on that of Roschelle and Teasley (1995). These authors
approached collaboration in terms of the construction of a ‘joint problem
space’, and oVered an analysis based on a single pair of how knowledge is
introduced into the emerging joint problem space, and how divergences
in meaning are monitored and ‘repaired’. We applied a similar approach
to four of the pairs from the present study, two ‘symmetrical’ and two
‘asymmetrical’ pairs (Joiner, Messer, Light and Littleton, 1994). The

symmetrical pairs selected made good progress, both in terms of the
performance of the pair in session two and in terms of the individual
performances of the pair members in session three. The asymmetrical
pairs performed poorly both as a pair and at post-test.
Four categories were used to code the transcripts of verbal interaction
in these pairs. The Wrst of these, ‘Repairs’, refers to instances where
conXicts of view are resolved through the use of justiWcations, counter-
suggestions or elaborations (this category is close to our earlier category of
negotiation). A second category concerned ‘Collaborative sentences’,
started by one partner and Wnished by another. Related to these are
‘Collaborative plans’, where one child starts a plan and the other com-
pletes it. Finally, ‘Simultaneous utterances’ are cases where the children
make essentially the same utterance simultaneously. The symmetrical/
successful pairs showed somewhat more of all of these types of verbal
interaction, but most notably they showed more than twice as many
Collaborative plans. Simultaneous utterances were present in the tran-
scripts of both symmetrical pairs (four in one and Wve in the other) but
were absent altogether in the asymmetrical pairs.
These results have to be treated with caution, given the limited sample
involved, but they closely echo those reported earlier (Wrst King and
Crown study) relating to shared decision making in successful pairs. Our
Wnding of superior performance and learning in symmetrical as compared
with asymmetrical pairings is consistent with the Wndings of some other
researchers (e.g. Whitelock, O’Shea, Taylor, Scanlan, Clark and O’Mal-
ley, 1993), but inconsistent with others (e.g. Howe, Tolmie, Anderson
and Mackenzie, 1992). It seems likely that the key factor is how far
success on the task depends upon prior knowledge that the participants
bring to the interaction.
However, the most striking result of this Wrst Honeybears study was
the negative one, namely that, overall, the children who worked in pairs
on the task at stage two were not signiWcantly better at post-test than
those children who worked alone throughout. Given that this result runs
directly contrary to what had been established in the Wrst King and
Crown study, we were obviously concerned to establish what made the
diVerence.
Though the software in use was diVerent, the task was formally isomor-
phic with the previous one, and the task remained a diYcult one. The lack
of diVerentiation of outcomes was certainly not accounted for by a ceiling
eVect. The pairs did do signiWcantly better at stage two, when actually
working together, though this eVect is a modest one which reaches
statistical signiWcance only because of the large numbers of participants in
this study. One hypothesis might be that the lack of stage three (post-test)
advantage for those who worked in pairs might have arisen because, by
contrast with the Wrst King and Crown study, the children in the present
study all worked alone at the Wrst (pre-test) stage. Thus the degree of
diVerentiation between the paired and individual conditions was less in
this study. It is also possible that the fact that children meet the task Wrst
on their own, and only later in pairs, alters the dynamics of the children’s
interaction with the task.
However, there is another design diVerence which oVers a possible
explanation for the diVerence in outcome, and one which lends itself
more readily to testing. This concerns the way in which the ‘individual’
sessions were handled. As noted earlier, whereas the children working
individually in the King and Crown studies were alone with an adult
experimenter and a computer, those in this study were always in a room
with three others classmates working on the same task. There was no
opportunity for overt interaction between the children during the session,
but might this simple fact of peer presence have made a diVerence? There
is some suggestive evidence on this point from a study of secondary
school physics students using a computer simulation (Whitelock, et al.,
1993). These authors report a facilitation eVect when students worked in
the presence of others, even without interaction, as compared with work-
ing entirely on their own. The next study was designed as a direct check
on whether this kind of facilitation was occurring here.
The second Honeybears study

A further study using the Honeybears software was set up to investigate
the possibility that peer presence, even without interaction, might be a
factor underlying the facilitatory eVect of pairing (Light, Littleton, Mes-
ser and Joiner, 1994). Thirty-two eleven year olds, sixteen boys and
sixteen girls, participated in the study. Half of the boys and half of the girls
were allocated to a ‘peer absence’ condition. These children were with-
drawn from the classroom individually, and were given a single thirty-
minute session working with the Honeybears task on their own, with only
the adult experimenter present. The other half (allocation being random)
were assigned to a ‘peer presence’ condition. These children were with-
drawn from the classroom in same-sex groups of four, and given a single
session on the Honeybears task working in the same room as one another
but without interaction. This second condition resembled that for the
Wrst Honeybears study, the machines being arranged so that the children
could not see one another’s screens. Care was taken to keep the form of
introduction and all the instructions constant across the two conditions.
The results of this study were striking. The mean performance level for
the children who worked in the presence of others (2.9) was signiWcantly
better than the performance of those who worked alone (1.8). Thus quite
apart from any consequences of productive verbal or practical interaction
between partners working together on the task, we have here evidence for
the operation of a quite diVerent kind of social facilitation of perform-
ance, which depends simply on the presence of classmates working on the
same task.
Overview
The three studies using the King and Crown software gave a more
encouraging picture of the beneWts of peer interaction than the studies
using standard problem-solving tasks reviewed in Chapter 2. With this
more open and extended type of problem solving (arguably more ‘face
valid’ in relation to educational experiences), the present studies seem to
show that both children and adults can reap considerable beneWt from the
presence of a collaborating partner. This peer advantage was evident even
though the children were given no more than a general invitation to work
collaboratively, without being forced into any particular pattern of inter-
action through the instructions or the structuring of the task itself.
The Wrst King and Crown study showed that the peer advantage could
carry over to individual subsequent performance on a slight variant of the
task. The second showed that the verbal interactions within the pair were
predictive of pair performance, of the subsequent performance of the
individual members of the pair, and indeed of the relative performance of
members of the pair at post-test. The third attested to the age indepen-
dence of at least some of the eVects observed, and suggested that the more
eVective use of available information to build a shared representation of
the task might be one factor explaining the advantage enjoyed by pairs.
The Wrst Honeybears study was to have been the deWnitive one. Its
scale, and full pre-test/post-test design, should have allowed a detailed
analysis of the basis of paired advantage. The observation that symmetri-
cal pairings were more eVective than other pairings seems to support an
interpretation couched broadly in terms of coconstruction rather than
conXict. However, other studies (e.g. Howe and Tolmie, 1999, using
tasks which focus on the judgements and prior knowledge that children
bring to physics tasks) have found evidence that asymmetrical pairs are
more eVective. We should acknowledge that diVerent psychological and
interactive processes are likely to be brought into play by diVerent types of
task, and we should not seek to draw overgeneralised conclusions. The
more important observation is that, as a whole, pairs did produce better
performance than individuals, but did not produce better learning, as

indexed by what the children could do by themselves at post-test.
The second Honeybears study lends support to an interpretation of this
result which focuses not on the paired but on the ‘individual’ condition. It
appears that the social setting of individual testing can have a marked
eVect on performance. The results echo those of Whitelock et al. (1993),
and on the basis of informal observation we concur with their observation
that it is not so much a case of facilitation of performance through the
presence of peers working on the same task, as a matter of degradation of
performance through anxiety when children work alone. Sitting as they
were at a novel computer, faced with a novel task under the watchful gaze
of an unfamiliar adult, it is perhaps not surprising that some of the
individuals appeared anxious and had diYculty engaging with the task.
From an experimental psychologist’s point of view, this type of individual
condition represents a natural and appropriate control condition against
which to assess the impact of social processes in the pairs. However,
this neglects the social and emotional dimensions of the control condi-
tion itself which, as evidenced here, have a considerable bearing upon
performance.
Though peer presence eVects and related phenomena are by no means
novel to social psychologists, their importance has been little recognised
by developmental and educational psychologists. Whilst, as we saw in
Chapter 1, there are diVerences between the Piagetian and Vygotskian
approaches, one of their similarities is that they are both essentially
cognitive in orientation. Neither has much to say about children’s aVec-
tive experience of the learning situation, or their motivation and self-
conWdence. To embrace the eVects demonstrated in this Wnal study, the
sense in which we normally understand terms such as interaction or
communication may need to be widened.
The next chapter will revisit the studies described in this chapter,
adding others using the same software, with a focus on gender as a factor
inXuencing children’s responses. Some of the wider issues raised by the
present studies, in terms of the more covert social processes that may be
operating, are taken up in Chapter 6.
5 Gender agendas
Boys, girls and computers

Much of the research on peer interaction processes in learning originating
in Piagetian and Vygotskian theories tends to be rather insensitive to the
‘non-cognitive’ characteristics of the individuals doing the learning. One
characteristic of learners that has drawn some attention, however, is
gender. This is especially true where the learning involves computers.
Psychologists and educationalists alike are well aware of the danger that
educational computer use might not just reXect, but actually amplify,
pre-existing sex diVerences (Light, 1997; Littleton, 1995, Littleton and
Bannert, in press).
Certainly girls often seem to be a good deal less enthusiastic about
computer use than boys. Surveys suggest that more girls than boys have
negative attitudes towards computers right across the school age range
(Martin, 1991; Robertson, Calder, Fung, Jones and O’Shea, 1995; Tod-
man and Dick, 1993, Whitley, 1997). A substantial proportion of both
boys and girls seem to regard use of the computer as being more ‘appropri-
ate’ for boys than for girls, and believe that boys like and use computers
more than girls do (Hoyles, 1988; Hughes, Brackenridge and MacLeod,
1987; Wilder, Mackie and Cooper, 1985). Girls may hold to the view that
‘girls in general’ are just as computer competent as boys, while at the same
time rating their own abilities lower than boys do (Shashaani, 1993).
An Australian study suggested that such sex diVerences in attitude
were particularly inXuenced by a sub-group of girls who were intensely
antagonistic to computers, while in general boys see computers as likely
to play a larger part in their future careers than girls do (Hattie and
Fitzgerald, 1988). International comparisons show that such sex diVeren-
ces in response to computers amongst students are widespread (Pelgrum
and Plomp, 1993; Janssen Reinen and Plomp, 1997). If anything, this
imbalance in response to computers has become stronger over the last
decade or so (Newton and Beck, 1993), and diVerences seem to become
more pronounced the longer the children are in school.
52
Gender agendas 53
The picture in respect of ‘voluntary’ usage of computers is perhaps less

clearcut, but not more cheering. Observational research suggests gender
diVerences in usage are already evident in primary schools (Straker,
1989) and that levels of participation in computer-related activities
amongst girls in UK secondary education tend to be low (Culley, 1988,
1993). This applied particularly strongly to optional computing activities
such as computer clubs, where in some studies 90 per cent of participants
have been found to be male. The proportion of girls entering for examin-
ations in computer studies and computer science is low and, if anything,
has tended to decline over time (Buckley and Smith, 1991; Culley, 1993;
Hughes, 1990). Certainly recruitment of girls to UK university courses in
computer science fell markedly during the 1980s (Hoyles, 1988). There
is also recent evidence that the proportion of female relative to male
Ph.D. students and lecturers in computer science departments is getting
progressively smaller (Camp, 1997).
There is evidence from other settings too that boys are more likely to
use computers than girls. Thus, for example, GreenWeld (1995) has
shown that in a science museum girls were more likely to use some of the
interactive displays (especially puzzles), but boys were more likely to use
those presented on computers. More importantly, home computers are
more likely to be bought for boys than for girls, and where they are
available, boys tend to use them more than girls for all purposes (Martin,
1991; Linnakylä, 1996). In one study it was found that ten times as many
boys as girls had sole access to a computer at home (Robertson et al.,
1995). It is possible that greater familiarity with computers out of school
contributes to the diVerences we see in school (Beynon, 1993). However,
Giacquinta, Bauer and Levin (1993) found only a modest amount of
general educational use of home computers (e.g. wordprocessing) and
almost no use of the home computer to support study of school subjects.
The authors point to a lack of liaison between home and school, and a
lack of willingness on the part of parents to get involved with the com-
puter, as being the main contributory factors. Home computers were
used mainly for game playing.
The ‘macho’ culture of computer games has often been commented
upon (Provenzo, 1991), but some of the same features in fact characterise
a good deal of educational software. Cooper, Hall and HuV (1990)
observed that the most popular software in use for teaching maths and
spelling employed the imagery of guns, missiles and warships, and adop-
ted a competitive, aggressive format. It is obviously possible that gender
stereotyping in the preferred formats of educational software might play a
part in shaping gendered responses to computers.
Sex diVerences in children’s preferred ‘style’ of working with com-
puters have been noted in various studies. Thus, for example, it has been
argued that while boys tend to adopt a more analytic and closed approach
in working with computers, girls tend to adopt a more open-ended and
exploratory approach (Turkle and Papert, 1990). Kirkup (1992) has
suggested that, whilst either style is compatible with using computers for
learning, software development for learning as well as for leisure has
tended to follow a masculine path.
Another stylistic diVerence sometimes reported points to a natural
linkage between the issue of sex diVerences in response to computers and
the issue of peer interaction in learning. Girls, it has been suggested,
prefer collaborative modes of working, while boys prefer to work alone
(Dalton, 1990; Hoyles, 1988). Hoyles, Sutherland and Healy (1991)
suggest that while boys seem to see interactions around the keyboard as
time-consuming diversions, girls see them as aVording opportunities for
mutual support and the development of ideas. Observations in mixed sex
classrooms suggest that, given the opportunity, boys tend to work individ-
ually at the computer, while girls tend to work cooperatively (Underwood
and Underwood, 1999). When children do work in groups around the
computer, girls tend to work more collaboratively than boys. The boys
tend to get fractious, one result of which is that they get more of the
teacher’s attention than the girls do (Culley, 1993).
Mixing the sexes does not seem to help. A substantial literature points
to the danger that boys will dominate mixed-sex interactions (Graddol
and Swann, 1989; Swann, 1997). The research literature more speciW-
cally on mixed-sex groups (mainly pairs) working with computers sug-
gests that these may indeed be relatively unproductive, with boys being
socially dominant in such situations (Siann, Durndell, McLeod and
Glissov, 1988). Watson refers to classroom observations such as: ‘Boys
came in Wrst and sat squarely in front of the screen. I had to remind them
to let girls have room to sit down’ (1997, p. 221).
Underwood, McCaVery and Underwood (1990) found that while
single-sex pairs did better than individuals when working on a sentence
completion task , the same was not true for mixed-sex pairs. Underwood,
Jindal and Underwood (1994) found that mixed-sex pairs of upper pri-
mary age children performed poorly on a computer-based language task,
showing little evidence of cooperative working, even given encouragement
to cooperate. Howe and Tolmie (1999) found with twelve to Wfteen year
olds working on a computer-based physics task that the amount of
task-related talk (expressed as number of dialogue turns) was substantially
lower in mixed-sex as compared to single-sex pairs. They also observed
that such interaction as there was in the mixed-sex pairs was not related to
learning outcomes in the way that interaction in single-sex pairs was.
Gender agendas 55
With larger groups of nine- to twelve-year-olds (consisting of three

boys and three girls) Pozzi, Healy and Hoyles (1993) did not Wnd signiW-
cant evidence of dominance. However, where groups fragmented, they
did so along gender lines, and antagonism was always across gender lines.
There are indications that larger mixed-sex groups may run very diVer-
ently depending on the balance of males and females in the group, with
equality or inequality of numbers being an important factor (Lee, 1993).
Evidence that girls actually learn less in the context of mixed-sex pairs
is not particularly consistent. Pheasy and Underwood (1994) found only
slightly lower levels of performance in mixed-sex as compared with
same-sex pairs. Hughes and colleagues (Hughes, Brackenridge, Bibby
and Greenhough, 1988; Hughes, Greenhough and Laing, 1992) found
no diVerence in learning outcomes between single and mixed-sex pairings
on a Logo task, except in one study where girls who had worked with boys
actually outperformed those who had worked with other girls. It seems
probable that the particular task involved has a good deal to do with the
results observed.
As Swann (1997) observes, the ‘boys dominate girls’ position is an
oversimpliWcation. What happens depends on the task and the situation
as well as on the participants. Fitzpatrick and Hardman (1994, and
Fitzpatrick, 1996) sought to establish how far any distinctive features of
mixed-sex task-focused interaction were speciWc to tasks involving com-
puters. They devised two similar tasks involving language puzzles, one on
a computer and one set within the context of a board game. Working with
seven and nine year olds, they found that mixed-sex pairs were marked by
signiWcantly more ‘assertive’ interactions than single-sex pairs.
This applied both with the non-computer and the computer-based
task. Interestingly, though, on the non-computer language task it was the
girl who was more likely to initiate an assertive interaction, whereas with
the computer task it was signiWcantly more often the boy who acted
assertively. Related to this was the observation that in the mixed-sex pairs
the girls handled the dice and counters more often than the boys, but
on the computer task the boys did the keying in more frequently than
the girls.
The evidence from both observational and experimental studies thus
suggests that boys exert a dominant inXuence in the context of mixed-sex
encounters with computers. It has been reported that girls in single-sex
schools have more positive attitudes to computers, and that in such
schools computer studies courses are popular and computer clubs thrive
(Culley, 1993; Gardner, McEwan and Curry, 1985). Taken together,
these Wndings have led to enthusiasm in some quarters for segregating
boys from girls in the context of computer use at school.
However, before leaving this review, it is important to note that there

appears to be little evidence that boys actually perform better on
computer-based tasks than girls, or that boys learn more than girls in
the context of computer-based learning. There are plenty of experimen-
tal studies which report no signiWcant diVerence in the achievements of
boys and girls on computer-based tasks, including, for example, some
of our own research on logic programming (Light and Colbourn,
1987). Likewise, although fewer girls take computer science courses
and degrees, those that do take such courses perform well (Lockheed,
Nielson and Stone, 1985). Thus the link between attitudes, choices,
inter-actions and learning outcomes looks likely to be far from straight-
forward.
In this chapter, we shall Wrst revisit the ‘King and Crown’ and ‘Honey-
bears’ studies described previously, attending to thus far unremarked sex
diVerences in the children’s performance. We shall then report some
further studies which directly address sex diVerences in children’s re-
sponse to these two tasks. Finally we shall broaden the range of tasks
under consideration and look at other aspects of gendered response to
collaborative learning with computers.
Kings, Crowns, Honeybears and gender

The second of the King and Crown studies described in the previous
chapter included a balance of boys and girls, with mixed-sex as well as
same-sex pairs. There were eleven boy–boy pairs, eleven girl–girl pairs
and eleven girl–boy pairs. The children worked as a pair in a Wrst session
and then individually (on a slight variant of the task) in a second session a
week later. They were also given a brief attitude questionnaire after the
second session. Seating positions and patterns of mouse control were
coded from videotapes of the Wrst (paired) session.
On the Wrst session the levels of success were modest (median perform-
ance level 2) and there was no signiWcant diVerence between the diVerent
types of pair. By the second session, when the children were working
individually, however, a large sex diVerence became apparent (see Fig-
ures 5.1 and 5.2). Figure 5.1 shows that at session two the boys (median
level six) were vastly better than the girls (median level two). Figure 5.2
shows the pattern according to pair type.
The children who had worked in boy–boy pairs did best, those who had
worked in girl–girl pairs did worst, with the mixed pairs intermediate. But
of course these last can be separated into boys and girls. When this is
done, it emerges that the boys in the mixed-sex pairs had done as well as
the boys in single-sex pairs, and the girls in mixed-sex pairs had done as
Gender agendas 57
Figure 5.1 Success in session 2 by gender
Figure 5.2 Success in session 2 by session 1 pair type

badly as those in the girl–girl pairs. We thus seem to see a substantial main
eVect for gender upon performance, but no eVect of mixed versus single-
sex pairing.
One of the post-experimental questionnaire items asked the children to
indicate on a Wve-point scale how much they agreed or disagreed with the
statement: ‘I like working on my own more than with a partner’. Regard-
less of pair type, the boys tended to agree with the statement, the girls
to disagree.
Seating positions were interesting. No attempt was made to regulate
this, but it transpired that in the mixed-sex pairs the boys were signiWcant-
ly more likely than the girls to end up sitting in a position (on the left)
which gave them right-handed access to the mouse. Whether this is
because the boys sought out this chair or because the girls avoided it we
are unable to say.
The children were given a general injunction to share the mouse, but
the evenness of ‘shares’ was variable. The girl–girl pairs showed the most
even distribution, the boy–boy pairs were intermediate, and the mixed-
sex pairs showed the least even distribution. The median number of
switches of mouse control between the members of girl–girl pairs in
the course of the twenty-Wve minute session was eleven. In the boy–boy
pairs the corresponding Wgure was four, and in the mixed pairs the
median was zero. In most of these pairs the boy controlled the mouse
throughout.
Thus there were quite marked sex diVerences in performance, and
quite marked pair-type diVerences in some non-verbal aspects of behav-
iour. However, our attempts to diVerentiate the pairs using the categories
of verbal exchange described in the previous chapter were not successful.
For example, the greater success of the boys was not associated with
signiWcantly higher levels of verbally explicit planning or negotiation.
These Wndings led us to consider the extent to which the noticeably
masculine characteristics of the King and Crown software might be
responsible for the gender-related diVerences in the apparent learning
outcomes. On the post-experimental questionnaire there was no indica-
tion that the girls were less attracted to the task, but this merely reXected
a ceiling eVect, in that all children tended to give very high ratings.
Objectively, the task consisted of an adventure game in the form of a
‘quest’ in which male characters engaged in stereotypically male activ-
ities. What would happen, we wondered, if we re-versioned the task so as
to avoid these features? The result was the Honeybears version of the
task, described in the last chapter. In attempting to produce a gender-
neutral version of the task we used images from fairy tales and advertise-
ments. Some commentators (e.g. Fitzpatrick, 1996) see the resulting
Gender agendas 59
scenario as distinctively feminine in orientation, but it was not our

intention to achieve this. Rather, we wanted something which was non-
masculine.
The Wrst Honeybears study, as described in the previous chapter,
aVorded an opportunity to look at both gender and pair-type eVects once
again with this new software. In the Wrst (pre-test) session all children
worked alone (or rather, they worked one-to-a-computer without interac-
tion, in groups of four). In the second session, the children were assigned
at random to boy–boy, girl–boy or girl–girl pairs, there being Wfteen of
each pair type. On a Wnal (post-test) session they worked on their own
again, as in the pre-test.
Figure 5.3 shows the pattern of performance of the three pair types
across the three sessions. The overall performance of the boy–boy pairs
was slightly better than that of the girl–girl pairs, with the mixed pairs
intermediate, but the diVerences both at session two (when the children
were working in pairs) and at session three (post-test) were not statisti-
cally signiWcant. None of the session one to three change scores revealed
signiWcant diVerences according to pair type.
We thus have a picture which is strikingly diVerent to that observed
with the King and Crown software. Gender diVerences are attenuated to
the point of being almost absent, even with a substantially larger sample
size. As before, though, pair type seems to make little diVerence in itself;
the pattern of performance across the three sessions for children who
work in mixed sex pairs in session two is very similar to that for children
who work in single-sex pairs.
We examined the pattern of mouse control in the mixed-sex pairs on
the second session to see whether the pattern of male domination seen
with the King and Crown software would be replicated here. It was not.
In fact the mean proportion of time the girls controlled the mouse (57 per
cent) was signiWcantly greater than that for which the boys controlled the
mouse (32 per cent). It being the second session, the children knew what
the task was going to be before they sat down. There was no sign at all of
the boys ending up sitting on the side which gave them ‘dominant hand’
access to the mouse; seating positions in the mixed-sex pairs were inde-
pendent of gender.
The results of the two studies discussed thus far seem to suggest that
superWcial gender stereotyping of the scenario within which a task is
presented may make a very substantial diVerence to the relative responses
of boys and girls. While it should not, perhaps, be surprising that superW-
cial features of the software will have some eVect, it also needs to be
remembered that many of the explanations put forward to account for
gender diVerences in relation to computers point to longer term factors.
Figure 5.3 The mean level of performance for the three diVerent pair
types
Thus, if gender diVerences are accounted for by diVerences in cognitive

style, preferred ways of working, or histories of exclusion, it should
surprise us that they turn out to be so labile.
However, though the diVerence between the versions of the software in
use is the obvious candidate to explain the diVering outcomes of these two
studies, it is by no means the only possible one. The studies diVered in
design in a number of ways, such as the inclusion of an individual pre-test
in the Honeybears study. Also, the fact that the individual condition in
the Honeybears study involved children working at separate machines in
the same room rather than on their own might be important. There is
some evidence that girls are more disadvantaged than boys by having to
work entirely on their own (Whitelock et al., 1993). Thus in order to
establish whether it was indeed the version of the software that was
making the diVerence, it was necessary to conduct some further studies
(Littleton, Light, Joiner, Messer and Barnes, 1998).
The software contrast studies

The Wrst software contrast study used the King and Crown and Honey-
bears versions of the software described in the previous chapter and used
in the two studies just described. The design was very simple. We took
Gender agendas 61
Wfty-two children, aged eleven, half boys and half girls. Half the boys and
half the girls were randomly allocated in advance to each version of the
task. The children were withdrawn from the class in groups of four (two
boys and two girls) and they went to a room in which there were four
computers. The two boys sat at one machine to start with, and were
introduced to one version of the task. They then moved to separate
machines and worked on that version of the task on their own. The two
girls, meanwhile, were introduced to the other version of the task by a
second adult, and went on to work on this version on their own. One of
the adult experimenters was male, the other female, and they balanced
their roles with respect both to the gender of the children and the version
of the software they were working with.
In the initial brieWng the children were given a tightly scripted tutorial
introduction in which they were shown the goal of the task. They were
also shown how to retrieve the information available in the software, how
to make a move, and the consequences of attempting to make an imposs-
ible move. Finally, the children were informed of the time available (thirty
minutes), but told not to worry if they did not solve the problem.
Figure 5.4 shows the average scores for the two gender groupings and
two versions of the task. Overall, performance was signiWcantly better on
the Honeybears than on the King and Crown version, but there was no
signiWcant eVect of gender. However, while the performance of the boys
remained virtually unaVected by the software type, the performance of
the girls was far superior when using the Honeybears software, resulting
in a signiWcant interaction eVect. Indeed, the girls’ mean level of perform-
ance on Honeybears slightly exceeded that of the boys, although this
diVerence is not statistically signiWcant.
The results clearly suggest that the performance of the girls was strong-
ly inXuenced by the version of the task employed. However, the study
suVered two weaknesses. Since the Honeybears version of the software
was not originally designed to make a comparison with the King and
Crown, it diVered in a number of ways, for example in the layout of the
various screens and the wording of the messages. These diVerences
potentially confound the present comparison. The facilities for automatic
computer recording of the children’s actions also diVered, precluding
detailed comparison of the task solution strategies adopted by the
children.
To deal with these problems, as well as to see how replicable the
previous result would prove, we designed a new version of the King and
Crown task called Pirates. This kept the storyline of the King and Crown
task, but it shared all design and interface characteristics with Honey-
bears. Using Pirates and Honeybears as the contrasting versions this time,
Figure 5.4 The mean level of performance for the girls and boys on the
King and Crown and Honeybears versions of the task
we took another sample of eleven year olds (forty-eight this time) and
re-ran the study just as before. Half the boys and half the girls were
randomly allocated in advance to Pirates and half to Honeybears.
The results are shown in Figure 5.5. This time there was no overall
eVect of software version, nor was there any overall eVect of gender. The
interaction between the two, however, remained signiWcant. As before,
the boys were apparently little aVected by which version of the task they
encountered, whereas the girls did far better with the Honeybears ver-
sion. On average, the girls outperformed the boys on Honeybears, though
this diVerence was not statistically signiWcant.
Inspection of the computer traces showed no signiWcant eVect of
gender or software type, nor any interaction between these, in terms of
how long it took the children to get started on the task, how much
information they searched before beginning to make moves, or in the
direction of initial moves. The only gender by software interaction dis-
covered from the trace concerned errors associated with the characters in
the scenario. Such errors might arise, for example, when a child failed to
Gender agendas 63
Figure 5.5 The mean level of performance for the girls and boys on the
Pirates and Honeybears versions of the task
specify which characters to take on a given move, or speciWed characters

not present at that location. With the Pirates version, girls made more
such errors than boys, whereas the reverse was the case for Honeybears.
More generally, it was apparent that many of the girls identiWed with
the characters, and afterwards, for example, some of them spontaneously
talked about which bear was their favourite. More than one talked about
taking particular bears on their journey to get the honey because ‘they
wouldn’t want to be left behind’. Such personiWcation of the characters
was not apparent amongst the boys, or with the King and Crown version.
In itself, identiWcation with the characters will not help to solve the task;
indeed, it could potentially hinder solution. However, it seems poss-
ible that this factor may have aVected the girls’ motivation to engage with
the task.
Taken together, the results of these two software contrast studies show
boys’ performance to be very little aVected by the version of the software
they were using. The girls’ performance, on the other hand, was very
substantially aVected by the version of the software they encountered.
While the King and Crown and Pirates tasks were not designed to appeal
particularly to boys, the roles of the characters were stereotypically mas-

culine ones and the means of transport (cars, ships and planes) had
masculine associations. The Honeybears version was designed to be free
of these male stereotyped elements; the characters had no explicit gender,
and the transports (pony, rowing boats and balloon) were less mechan-
ical. Nonetheless, the underlying task was exactly the same for both
versions.
It is possible that the relatively high lability of girls’ performance is a
function of lower experience/conWdence with computers rather than of
gender as such. Robinson-Staveley and Cooper (1990) found with older
women students that the performance of women with low familiarity and
conWdence with computers is greatly aVected by subtle manipulations of
their expectations about how well they will do, whereas women with
higher levels of experience of computers are largely unaVected by such
manipulations. Unfortunately we do not have independent evidence with
the present samples of whether the boys were generally more experienced
and conWdent with computers than the girls, but other research men-
tioned earlier suggests that this is likely.
These two software contrast studies oVer an important message to
those involved in designing or choosing software for children’s use.
Political correctness may be out of fashion, but here there is clear evi-
dence that unthinking gender stereotyping of computer-based learning
materials really can make a diVerence to children’s learning. Given an
initial motivation to engage with the task, the girls were adept at handling
the interface and thinking their way through the problem. In the following
section, though, we want to move on from consideration of software
eVects to more ‘interactive’ aspects of the gender agenda. To begin with
this will involve a return to the other aspect of gender addressed in the
earlier studies, namely the relative success of single-sex as compared to
mixed-sex pairings.
Interaction and ‘coaction’ in mixed gender pairs

In the Wrst section of this chapter, we referred to studies which suggest
that boys tend to dominate computer resources, get the best machines,
keep girls out, and so on. There is a strong suggestion in the literature that
the presence of boys may be deleterious to the performance of girls in
this area.
The Wrst King and Crown study, as we have seen, did produce evidence
of this kind of domination by boys of the interaction with the computer.
On the other hand, even in that study girls in the mixed-sex pairs seemed
to gain as much from the interaction session as did girls in girl–girl pairs.
Gender agendas 65
More generally, the evidence that mixed-sex pairs and single-sex pairs are
marked by diVerent styles of interaction is much clearer than the evidence
that children in mixed-sex pairs perform less well or learn less.
Fitzpatrick (1996), for example, showed in a number of studies that
interactional patterns in mixed-sex pairs were characterised by con-
strained interaction, greater task demarcation and less collaborative talk.
However, the girls in these pairs showed no less progress in understand-
ing at post-test than girls who had worked in single-sex pairs. Such results
should caution us to recognise that the relationship between interaction
and learning is far from straightforward.
We saw in the previous chapter (Wrst and second Honeybears studies;
Light et al., 1994) that children tended to perform better on the Honey-
bears task when another child was present in the room working on a
similar task than when they were alone with the experimenter. In those
studies the children present were always of the same gender. We went
on to consider whether mixed-sex pairing might make a diVerence even
here, where there was not overt interaction (we called this a ‘coaction’
condition).
Sixty-two eleven-year-old children participated in our Wrst study of this
issue (Light, Littleton, Bale, Messer and Joiner, in press). The children
were assigned at random to ten boy–boy pairs, ten girl–girl pairs and
eleven mixed-sex pairs. They were taken in their assigned pairs from their
classroom to another room in the school by an unfamiliar (male) experi-
menter, who gave a brief introduction and demonstration of the Honey-
bears task and then sat each of them at a computer. The children were
told that they probably would not Wnish the task before the end of the
session, but that they should see how far they could get.
The two computers were close to one another, but arranged so that the
children could not see one another’s screens. They could see one an-
other’s faces, but were asked not to talk to one another during the twenty
minutes they spent working on the task. The adult remained present
throughout the session, but did not intervene in any way. The study
involved only a single session for each child.
Figure 5.6 shows the pattern of performance. Boys did slightly better
than girls overall, but more interestingly there was a signiWcant gender by
pair type interaction. Boys tended to do better when their ‘partner’ was a
girl, but girls tended to do less well when their ‘partner’ was a boy. The
trace data reveal similar statistical interactions in respect of information
searching. Girls in mixed-sex pairs spend less time searching for informa-
tion (both before the Wrst move and overall) than girls in single-sex pairs,
whereas the boys spent more time in information searching than boys in
single-sex pairs. Thus it seems that in the mixed-sex pairings, the boys are
Figure 5.6 The mean level of performance for the mixed and single-sex
dyads
behaving more planfully, but the girls less planfully than in single-sex
pairings.
The surprising feature of these results is that, whereas in previous
studies children in mixed-sex pairs actually performed at very much the
same level as those in single-sex pairs, here we are seeing a signiWcant
eVect of pair type on performance. But in this case, unlike those which
have gone before, the children are not working collaboratively at all, but
simply working in proximity with one another. Thus despite evidence
that in interactive mixed-sex pairs the boys do indeed tend to dominate
the interaction with the computer, we Wnd clearer evidence for the eVect
of partner gender on performance when partners do not interact around
the task than when they do.
On a post-experiment questionnaire, the children were asked whether
they thought boys or girls were most interested in, and better at using,
computers, or whether they thought both were equal. About half of the
children (boys and girls equally) thought that boys were more interested
in computers than girls, whereas no child thought the opposite. Only
about one third of the children (boys and girls equally) thought that boys
were better at using computers than girls, but again no child thought the
opposite. Thus we can say that between a third and a half of the children
Gender agendas 67
did associate enthusiasm for, and expertise with, computers more

with boys than with girls. It seems a fair assumption that this perception
has a good deal to do with the observed eVects of partner gender on
performance.
Before attempting to unpick this further, we decided to conduct a
larger study which would allow direct comparison of ‘interactive’ and
‘coactive’ mixed-sex pairings. Instead of a single session study, we used a
full three stage design, with pre-test and post-test separated by a longer
(forty minute) session in which the children worked in pairs either inter-
actively or coactively.
In the Wrst (‘pre-test’) phase, ninety-six eleven-year-old children were
withdrawn from class in single-sex pairs from their classroom by a female
experimenter. After a brief, scripted introduction, each of the children
worked separately at a computer for ten minutes on a further variant of
the problem-solving software called ‘Princesses’ (in which three prin-
cesses have to retrieve the queen’s crown, avoiding a sea-witch).
On the basis of their performance on this pre-test, the children were
assigned to groups and conditions. All pairs comprised children from
diVerent classes in the same school who had the same pre-test score. By
taking children from diVerent classes we hoped to minimise the speciWc
prior knowledge they would have of one another’s interests and abilities.
Half of the children (eight boy–boy pairs, eight girl–girl pairs and eight
girl–boy pairs were assigned to the interactive condition. The other half
were assigned to the coactive condition. All sub-groups were balanced in
terms of mean pre-test scores.
A week after the pre-tests, the children were brought to the testing room
with their assigned partner, and given an introduction to the Honeybears
version of the task. They were told that they would have forty minutes to
work on the task, but not to worry too much about getting it Wnished
because they would have another go on their own the following week.
The children in the interactive condition were seated at a single com-
puter, and were encouraged to share the mouse and to help each other.
The children in the coactive condition each sat at their own computer, the
screens arranged so that they could not see each other’s, and they were
asked not to talk to each other during the session. The Wfteen-minute
post-test session a week later ran just as the pre-test, using Honeybears, but
with alterations to some locations so that the children had to ‘re-solve’ the
problem rather than just reproducing a remembered sequence of moves.
Since the pairs and sub-groups were all balanced in terms of pre-test
scores, the results can be given simply in terms of performance at post-
test. Figure 5.7 shows the outcome for the interaction condition. There is
no overall sex diVerence, nor, more importantly for present purposes, is
pairs for the interacting pairs
there any diVerence as a function of sex of partner. Children who have

worked interactively with an opposite-sex partner perform in just the
same way at post-test as children who have interacted with a same-sex
partner. This replicates our previous Wndings with interactive pairs.
Figure 5.8 shows the outcome for the coaction condition. We see the
same gender polarisation in the mixed gender dyads that we saw in the
study just described, reXected once again as a statistically signiWcant
interaction. Girls who had worked in the presence of a boy in the practice
session did less well at post-test than those who had worked in the
presence of another girl. Boys who had worked in the presence of a girl
did better than those who had worked in the presence of another boy.
The children’s reponses to the question posed after the Wrst of these
studies conWrmed the general Wnding that a substantial proportion of the
children (both boys and girls) expect boys to be better with computers
than girls. Actually, as we have seen over a series of studies with the
Honeybears task, there is little or no diVerence in the average perform-
ance of boys and girls. Indeed, in the study just described, pre-test scores
were used to create matched pairs, so that the girls and boys were equal in
terms of initial ability on the task.
Gender agendas 69
pairs for the coacting pairs
Where the children work collaboratively, as in the interaction condi-

tion, they will have every opportunity to discover each other’s competen-
cies and incompetences. This might well result in a defusing of any
gender-stereotyped expectations they may have had. In fact, as we saw in
studies reported earlier in this chapter, the girls may actually dominate
the interaction, at least in terms of executive control of the computer.
Where the children were working in close proximity but without inter-
action, as in the coaction condition, they did not have the same opportun-
ity to defuse any gendered expectations that they might have brought to
the situation. They were kept very much aware of one another’s presence;
they could see each other whenever they looked up, and could hear the
‘beeping’ of one another’s machines in response to keyboard entries.
These noises were not informative as to the partner’s progress, but they
may well have expected that their relative degrees of success on the task
would become evident as the session went on. In fact this was not the case
unless one of them completed the task inside the session time, which was
infrequent.
It seems plausible to suppose, then, that in the mixed-sex situation the
boys were particularly challenged to achieve well, while girls may have
been inhibited by seeing themselves to be at a disadvantage from the

outset. This explanation links gendered perceptions of relative compet-
ence to the eVects observed. In this connection, mention should be made
of one last Honeybears study, which, unlike the others, drew boys and
girls from independent single-sex schools (Flynn, 1995).
The children were aged ten to eleven, and had been privately educated
in single-sex preparatory schools for most or all of their schooling. Gender
stereotyping in relation to the curriculum is reported to be much less
pronounced in single-sex as against mixed-sex schools (Bone, 1983), and
indeed it transpired that these children showed no signiWcant sex diVer-
ence in enthusiasm for computers as indexed by a standard questionnaire
(Todman and File, 1990).
Ten boy–boy pairs, ten girl–boy pairs and ten girl–girl pairs were
established at random, drawing from two schools. They were given a
single twenty minute session on Honeybears, working ‘coactively’ at two
computers. There was no trace of the gender polarisation in the mixed-
sex pairs that we had seen in the previous studies. Both the girls and the
boys in the mixed-sex pairs behaved as the boy members of such pairs had
done in the Wrst of the studies described in this section, that is, they spent
signiWcantly more time on information searching than did children in the
single-sex pairs.
Overview
In the introduction to this chapter we provided a brief review of the now
quite extensive research literature focusing on gender diVerence in re-
sponse to computers. Such gender diVerences are all seen as working to
the disadvantage of girls. The pattern of gender diVerence observed with
our King and Crown task seemed to conWrm the gloomy picture; the boys
dominated the mixed-sex interactions and performed signiWcantly better
than the girls on the task at post-test. Nonetheless, it is important to note
that the extent of the gender diVerence in performance was independent
of whether the children had worked in same-sex or mixed-sex pairs.
The introduction of the Honeybears version of the task was marked by
a striking reduction in both the gender diVerence in performance and the
dominance of the computer interface by the boys. Controlled contrasts
of the two versions of the task showed quite clearly that, even with
exactly the same underlying task, the setting of the task made a great
deal of diVerence to the performance of the girls. The gender diVerence
which was so striking in the King and Crown task could be ameliorated
or indeed eliminated by simply altering the storyline and characters
involved. Whether versions of the task could be found which create
Gender agendas 71
signiWcant performance advantages for the girls remains to be seen, but

the evidence presented here seems suYcient to merit the close attention
of those involved in educational software development and selection.
The software contrast studies strongly suggest that there is nothing
immutable about gender diVerences in this area. The results with mixed-
sex pairs in interaction around the computer suggest that even where
there is apparently a high degree of dominance, the relationship to learn-
ing outcomes is by no means straightforward. The Wnal studies described
in this chapter contrasted mixed-sex pairs in interaction with mixed-sex
pairs whose members were simply ‘coacting’: working on separate com-
puters without (overt) interaction. The degree of gender polarisation of
performance was stronger in the latter case than in the former.
Our tentative interpretation is that it is not gender so much as self-
conWdence that is at issue. This interpretation Wnds some support in a
study by Robinson-Staveley and Cooper (1990), mentioned earlier.
When expectation of success on a computer task was artiWcially manipu-
lated, both male and female students with low induced expectancy of
success were impaired by the presence of someone else, while those with
high induced expectations performed better in the presence of someone
else than when alone.
Some further evidence comes from a quite separate study using a
computer based mathematical reasoning task (Joiner, Messer, Steele,
Light and Littleton, 1993). In this study, again with children aged ten and
eleven, we found that the performance of children with high computer
experience was facilitated by the presence of others working on the same
task, whereas the performance of children with low computer experience
was inhibited by the presence of others, both in terms of the time taken
and the number of problems successfully solved.
An interesting Wnding relating directly to software contrasts is reported
by Cooper, Hall and HuV (1990). Their study was based upon a com-
parison of maths software with very masculine imagery (called ‘Demoli-
tion division’, involving tanks etc.) with a comparable maths package
without any such aggressive imagery. The children (twelve- to fourteen-
year-olds) completed an anxiety inventory after working on one or other
type of software. When working in mixed gender ‘coactive’ grouping of
sixteen in the school computer room, the girls reported higher levels of
anxiety after working with the ‘aggressive’ software than with the other,
and the reverse was true for the boys. This interaction between the sex of
the child and the software type was not found when the children used the
software on their own, in privacy.
Unfortunately, Cooper et al.’s study did not include performance
measures for both types of software, and their demands were in any case
not directly comparable. Nonetheless, their conclusion that: ‘Gender

biased software is a stressor to the opposite sex if it is encountered in the
presence of observers or possible observers’ (1990, p. 426) is relevant to
interpreting the present results.
Overall, the conclusion to be drawn from the studies reported in this
chapter must be that the eVectiveness of peer interaction in learning
situations may be as much a matter of social comparison as it is of social
interaction. The focus of both Piagetian and Vygotskian work in this Weld
has been on identifying the cognitively constructive elements of interac-
tion. The results discussed here suggest that a more social-psychological
approach, focusing on children’s perceptions of their own ability relative
to the task and to their partner, may be at least as illuminating. Thus we
are pushed from an interpretation of peer facilitation eVects in terms of
observable, cognitively productive forms of interaction towards a recogni-
tion that at least some of the eVects may depend on less directly observ-
able social processes. These will be the focus of the next chapter.
6 Social comparison and learning
Social comparison and ‘social loaWng’

The study of processes of social comparison has been central to social
psychology but marginal to developmental psychology. However, as we
have seen in the last two chapters, results of peer interaction studies seem
to force issues of social comparison to our attention. This chapter begins
with a brief overview of some of the relevant background literature, with a
particular focus on recent research by Monteil, Huguet and colleagues in
France. We shall then introduce further empirical studies which build
upon Monteil’s work as well as our own.
Social comparison processes bridge the gap between issues of self
and the sense of personal identity on the one hand, and of interper-
sonal and intergroup processes on the other. At its baldest, the term
social comparison simply refers to any comparison an individual might
make between his or her own attributes or abilities with those of some-
one else. Festinger (1954) highlighted the power of social comparisons
in situations where individuals have no Wrm objective criterion by
which to judge themselves or their performance. Many of the judge-
ments we make about ourselves are relative, and may Xuctuate accord-
ing to who else is available to compare ourselves with (Durkin, 1995;
Gergen, 1977).
The role of social comparison processes in socialisation (in the acquisi-
tion of norms and evaluation of the self against those norms) has been the
subject of some developmental research (Ruble, 1983). Seven to eleven
year olds given repeated sessions on a mathematics task were found by
Ruble and Flett (1988) to pay more attention to comparisons of their own
performance to that of peers than to comparisons of their own present
and previous performance, though the more able older children did show
greater interest in the latter type of comparison. Durkin (1995) suggests
that with development, individuals make more and more selective com-
parisons, based on such things as perceived similarity and common group
membership.
73
Social comparisons begin early in life, but it appears that comparisons

of ability Wgure prominently in children’s thinking only from about the
age of seven (Ruble and Frey, 1991). It is diYcult to be sure how far
schooling, as opposed to age, might be responsible here. There is some
evidence that social context can make a signiWcant diVerence. For
example, seven- and eight-year-old Israeli children raised in kibbutzim
are less likely to think in terms of competitive, ability-based comparisons
than their urban peers are (Butler and Ruzany, 1993).
Social comparison eVects in school contexts are particularly relevant
for present purposes. Marshall and Weinstein (1984) showed that a
variety of contextual factors could make a diVerence to the way social
comparison factors aVected school performance, notably the ‘visibility’ of
the social comparisons themselves. The eVects of performance-related
feedback, for example, diVered according to whether that feedback was
delivered to the children privately or in front of the group. Not surprising-
ly, it appears that cooperative working arrangements weaken the focus on
social comparisons of individual ability, whereas competitive working
arrangements heighten the salience of such comparisons (Ames, 1981).
A considerable body of research on the eVects of evaluative feedback on
performance indicates that such eVects are most frequently positive, but
sometimes negative (Kluger and DeNisi, 1996). Monteil and Huguet (in
press) believe that the apparently inconsistent results can be explained if
attention is given to the circumstances of feedback. In particular, the key,
as they see it, is whether feedback on performance is given in circumstan-
ces which place the child in a situation of social comparison with another.
Monteil and Huguet have returned to some of the earliest studies of
social facilitation of learning to try to establish a role for social compari-
son. Triplett (1898) is widely cited as oVering a clear demonstration that,
for children aged nine to thirteen, the presence of another child working
on the same task (a simple motor skill task) facilitated performance.
Actually, according to Monteil and Huguet, a quarter of Triplett’s par-
ticipants showed no eVect and another quarter showed a negative (inhibi-
tory) eVect of peer presence.
Burnham (1905) compared the performance of children in classes at
school with the performance of children working alone at home, and
concluded that performance in class was superior for activities which did
not demand much originality, but the opposite was true for more creative
tasks. Much later, Zajonc (1965) also oVered evidence which seemed to
point to a task eVect, namely that peer presence is facilitatory with simple
tasks but inhibitory with complex tasks.
Zajonc oVered an account of his results in terms of an essentially
behaviourist ‘drive theory’, according to which the presence of others
Social comparison and learning 75
raised the level of general drive, and favoured the production of the most
readily available response to the task in hand. With simple tasks, this
response is likely to be correct, so that performance facilitation results.
With more complex tasks, successful solution may depend on the inhibi-
tion of such dominant responses, and raised levels of general drive may
thus reduce the quality of performance.
Cottrell, Wack, Sekerak and Rittle (1968) argued that it was not so
much the mere presence of individuals that makes a diVerence, but rather
the ‘evaluation potential’ of the audience. They showed, for example, that
for individuals performing a task, the presence of others made no diVer-
ence if they were blindfolded, but it did make a diVerence if they were
watching the performance keenly.
Both the pattern of Wndings reported by Zajonc and his approach to
explaining them have found mixed support in subsequent research (Bond
and Titus, 1983; Glaser, 1982). The predicted pattern of results for
simple and complex tasks does not always occur, especially in relation to
tasks involving memory and cognition (Hartwick and Nagao, 1990).
There have been a number of attempts to reconceptualise the issue of
social facilitation in more cognitive terms. For example, Manstead and
Semin (1980) argued that the presence of the other could be seen as
making demands on Wnite attentional resources, to the detriment of
learning tasks calling for considerable attention. ‘Automated’ performan-
ces, on the other hand, demand little attention and may beneWt from
anything which maintains alertness.
This focus on attentional resources and their disposition underpins the
approach of Monteil and Huguet (in press), who report an unpublished
study in which Huguet, Monteil and Galvaing used the Stroop test to try
to distinguish dominant response explanations from explanations based
on attention. With, for example, the word (e.g. ‘red’) written in a diVerent
colour (e.g. green), the participant is asked to name the colour. The
correct answer is thus ‘green’, but the interfering dominant response (to
read the word) will tend to generate the erroneous response ‘red’.
In these circumstances, if the presence of another person increases the
tendency to generate the dominant response it should lead to a decrement
in performance. However, if the presence of another person makes it
harder to focus attention, as Monteil and Huguet suppose, then it might
be expected to improve performance. The results showed that provided
the other person was attentive and looking at the participant, Stroop
interference was reduced and performance improved.
But Monteil and Huguet are seeking to foster a position which is not
only more cognitively oriented than Zajonc’s, but also more socially
oriented. They emphasise the importance of the participant’s personal
perspective on the task and the broader situation within which it is set.
Participants come to learning situations with a history of experience of
their performance across a range of more or less related tasks in more or
less related situations; an ‘autobiography’ of themselves as learners.
The central point here is that the presence of another can generate
diVerent signiWcations according to the history of the individual learner.
Social contexts do not exist independently of the individual. Rather:
‘Social contexts exist only through the intervention of cognitive structures
of contextualisation, such as those linked to the autobiographical memory
of the individual’ (Monteil and Huguet, in press). In other words, while
individual performance depends upon social contexts, social contexts can
only be understood in terms of individuals.
Monteil and Huguet (in press) point out that a number of studies over
the last twenty years or so have suggested that the experience of the
individual immediately prior to the peer presence situation can make a
signiWcant diVerence to the outcome observed. For example, Mash and
Hedley (1975) showed that a positive interchange with the experimenter
led to facilitative eVects of subsequent peer presence on a simple motor
task, but a negative interchange with the experimenter resulted in inhibi-
tory eVects of subsequent peer presence. Similarly, Seta and Hassan
(1980) showed that audience eVects could vary according to the individ-
ual’s immediately prior history of success and failure on the task. Using a
memory task, they showed that the presence of an audience improved the
performance of those with a previous history of success but impaired that
of those who had encountered failure.
If social comparison is one key concept in twentieth century social
psychology, ‘social loaWng’ is another. Ringelmann (1913) observed that
the total physical eVort produced by a group (in response to a task
involving pulling on a rope) was less than the sum of the eVort put in if
each member of the group worked alone. The diVerence was found to be
proportional to the size of the group. The term ‘social loaWng’ was
subsequently introduced to refer to the tendency of individuals to pro-
duce less eVort in group than in individual situations. It has been shown
that social loaWng does not occur in a situation where the individual
contributions of the participants are expected to be identiWable (Will-
iams, Harkins and Latane, 1981). It seems that evaluation (or the poten-
tial for evaluation) of relative performance is a determining factor in this
regard.
Monteil and Huguet (in press) are critical of research in this area
because of its characteristic neglect of the history of experience of the
individual participant. They cite a study by Sanna (1992) in which
individuals were paired for a vigilance task. Having received feedback
indicating that they were doing better or worse than their partner, they
went on either to work as a pair (where their success would be judged on
‘group product’) or coactively (where their individual success levels
would be visible). Those who had been favoured by the initial compari-
son subsequently performed better in the coaction than the group condi-
tion (a result consistent with social loaWng). However, the reverse was
true of those whose initial feedback had suggested that they were doing
less well.
It seems, then, that social comparisons can moderate the eVects at-
tributed to social loaWng. So too can individual diVerences of a more
general/stable kind. Huguet, Charbonnier and Monteil (1995) showed
that reduced productivity in a collective as against a coactive situation was
characteristic only of students who were high on ‘self-uniqueness’, as
indexed by a separately administered questionnaire. Extending this to
cultural diVerences, Gabrenya, Wang and Latane (1985) have shown in a
comparison of Chinese and North American students that social loaWng
was characteristic only of North American males. Chinese males showed
exactly the opposite pattern (‘social striving’). Without wanting here to
open up the complex debate about individualistic and collectivist cul-
tures, we can simply recognise that universal accounts which take no
account of individual, social and cultural characteristics are unlikely to be
adequate to the phenomena. For example, social loaWng appears to be
more a male than a female phenomenon: Karau and Williams, 1993, and
Monteil and Huguet, in press, report data indicating that while boys
produce better memory recall performance in ‘public’ than in anonymous
conditions, the reverse is true for girls.
Social comparison and children’s learning

Monteil (1988) describes a series of studies which attest to the subtle
interaction of individual abilities, personal histories and social contexts
for learning. In his Wrst study, participants were fourteen- and Wfteen-
year-olds, selected as having either high or low levels of general educa-
tional attainment. The children were placed in groups of eight (four high
performers and four low in each group). Then, in the ‘Social comparison’
condition, the children were publicly told their actual educational levels.
In the ‘No social comparison’ condition they were told that they were all
at the same (unspeciWed) level. They were then taught some biology in a
forty-Wve minute lesson, either in a ‘individuated’ condition (where they
were told to expect to be questioned individually during the lesson), or in
an ‘anonymous’ condition (where they knew that they would not be
singled out for questioning). No questioning actually occurred in either
condition, but after the lesson all students completed a written test on the
material taught.
In the ‘No social comparison’ condition, the high attainers did better
on the test than the low attainers, and the individuation/anonymity
manipulation had no eVect. In the ‘Social comparison’ condition, how-
ever, individuation led to an ampliWcation of the diVerence between the
high and low attainers, while anonymity reduced the diVerence to almost
nothing. Given the combination of social comparison and anonymity, the
high attaining pupils did rather less well than they had done in other
situations, while the low attainers did exceptionally well.
Similar results were obtained in a second study conducted with only
high attaining pupils. Here, social comparisons were introduced by giving
pupils feedback on a prior task, indicating success or failure. Even when
this feedback was actually uncorrelated with real performance, it engen-
dered the same pattern of results. Negative feedback was followed by
better learning under anonymous than under individuated conditions,
while positive feedback was followed by better performance under indi-
viduated than under anonymous conditions (Monteil, 1988).
Subsequent studies showed that the eVects shown in these initial
studies are also conditioned by the academic context in which the task is
set, being evident for highly valued disciplines such as mathematics and
biology, but not for lower valued disciplines such as ‘technical and
manual education’. This observation opened up a new line of research, in
which Monteil and Huguet (1991) explored the eVects of representing a
given standard task to some students as ‘geometry’ and to others as
‘drawing’. The actual task was always the same and involved a complex
rectilinear Wgure (see Figure. 6.1), which the pupils were shown for a
short period of time and asked to remember. Their success on both a
recall (which involved drawing from memory) and a recognition task was
shown to vary as a function of both discipline context and their own
academic standing. When the task was represented as being ‘geometry’,
there was a large diVerence in memory performance between the children
classed as generally high attainers and those classed as low attainers,
with the high attainers doing better. However, when the task was repre-
sented as ‘drawing’ there was no diVerence in performance between the
two groups.
This intriguing result prompted us to attempt a replication in the UK.
It seemed possible that the results reported by Monteil and Huguet might
depend upon certain features of the French education system which
diVerentiated it from the British system. In particular, it appears that
pupils are made more explicitly aware of their own levels of academic
performance in France than in the UK. For example, they are required to
Figure 6.1 The complex rectilinear Wgure
repeat an academic year if they fail to reach a certain standard. Similarly,

it seemed to us possible that the high valuation of some disciplines
(especially mathematics) might be more characteristic of French than of
British schools.
For reasons of convenience, we worked with children of middle school
age (eleven- and twelve-year-olds) rather than the somewhat older sec-
ondary pupils used by Monteil and Huguet. We used the terms ‘maths’
and ‘art’ (in place of ‘geometry’ and ‘drawing’) in relation to the same
rectilinear Wgure illustrated in Figure 6.1 (Light, 1994; Littleton, Light,
Robertson and Beeton, in preparation).
A pilot study involved the presentation of the memory task to Wfty-two
children, unselected for ability, to see whether the representation of the
task as maths or drawing made a signiWcant diVerence for British children
of this age level. The children were taken from their classes in groups of
four and were told that we were doing a survey in diVerent schools of how
well children of their age could do in diVerent subjects. Half of the groups
were then told that they had been chosen for a maths test and the other
half that they had been chosen for an art test. Each child, working
individually, was presented with the Wgure on paper to look at for one
minute. They were then given a pencil and paper and asked to reproduce
it as well as they possibly could, as a freehand drawing. Scoring of

accuracy of reproduction was done blind by judges, using a scoring
system (based on the number of elements present and correctly posi-
tioned) which showed good interjudge reliability.
The accuracy of recall scores did show a diVerence according to condi-
tion of presentation. The children who had been told that the task was to
test their ability in maths obtained signiWcantly higher scores (mean 22)
than those who were told that it was to test their ability in art (mean 17).
We were unable to discern any diVerences in the style or technique
adopted by the children working under the two presentation conditions,
but our impression was that the children were working on their reproduc-
tions for longer in the maths condition than in the art condition.
For the main study we worked with forty-eight eleven and twelve year
olds, selected as representing particularly high and low levels of general
educational attainment. At our request, teachers of six state middle
school classes selected the two highest attaining and the two lowest attain-
ing children of each sex from their class, giving us twenty-four high
attainers and twenty-four low attainers. Procedure was much as before,
except that the children were taken from their class individually. On
arrival in the testing room, they were told that they were taking part in a
survey, as in the previous study. However, before they were told which
subject they were going to be tested in they were asked to rank nine school
subjects in terms of how important they were, and then in terms of how
good they thought they were at each. After this, half of them were told
that they were going to take part in a maths test, the other half that they
were going to take part in an art test.
Procedure was then as before, except that while they worked on repro-
ducing the Wgure they were timed, and they were asked to stop when they
could not remember any more. The pattern of results echoed that ob-
served by Monteil and Huguet (1991), in that the high attainers produced
signiWcantly better recall performance when they were told that it was a
maths test (mean score 23) than when they were told that it was an art test
(mean 19). The reverse was true for the low attainers (mean 13 for maths,
15 for art), though this diVerence was nonsigniWcant.
The same pattern was true to a more marked degree in respect of the
time spent by the children working on their reproduction of the Wgure.
High attainers tended to be quicker overall, but spent more time on recall
in the maths condition (mean 121 secs) than in the art condition (mean
115 secs). By contrast, low attainers spent more time on recall in the art
condition (mean 143 secs) than in the maths condition (mean 115 secs).
This resulted in a statistically signiWcant interaction between attainment
level and presentation condition.
For the high attainers, the modal ranking of importance for maths was
1 (i.e. it was seen as the most important subject of the nine subjects
listed). For the low attainers the modal ranking of the importance of
maths was signiWcantly lower, at sixth in the list of nine. However, both
high and low attainers put art lower down the importance ranking than
maths, with a modal ranking of 8 from both groups. For the high at-
tainers, maths was seen as signiWcantly more important than art, whereas
for the low attainers this diVerence was not signiWcant.
Overall, the higher the relative importance children attached to the
subject in which they were supposedly being tested, the longer they
tended to spend on the reproduction of the Wgure from memory. This
relationship did not quite reach a satisfactory level of statistical signiW-
cance, but again seems to indicate that taking a longer time is indicative of
greater eVort and attention.
When asked to rank the subjects in terms of how good they thought
they were at them, the high attainers gave themselves a modal rank of 2
for maths, whereas the lower attainers gave themselves a modal rank of 9
for maths, indicating that this was the subject they were least good at.
However, there was a great deal of individual variability, and this diVer-
ence was not reliable. For art, however, there was a statistically reliable
diVerence between the rankings the low attainers gave themselves (modal
rank 1) and those the high attainers gave themselves (modal rank 9).
Taken together with Monteil and Huguet’s earlier Wndings, these
results testify to the extent to which the apparent disciplinary context of a
task can make a diVerence even when the task itself remains exactly the
same. The success children have with academic tasks clearly can depend
upon how the task is labelled and how the children perceive their own
abilities in relation to that label. More generally, the interaction between
levels of attainment and disciplinary context shows once again that given
contexts depend for their eVects upon individual attributes and histories.
In an extension of this work with the memory task, we returned to the
question of learning with computers, and to the issue of gender. The
Wgure shown in Figure 6.1 was presented to sixty-four eleven year olds,
half of them male, half female. To half of the boys and half of the girls, the
Wgure was presented as before, on paper. To the other half it was pres-
ented on a computer screen. In all cases, the children were required
simply to look at it, and then to reproduce it from memory on paper. Thus
even in the ‘computer’ condition, the children did not actually have to do
anything with the computer at all.
Children came to the testing room in same-sex pairs, but worked
separately, without interaction and without being able to see what each
other were doing. Either both were in the ‘computer’ condition, or
Figure 6.2 The mean recall score for the girls and boys for the two
modes of task presentation
neither were. If they were in the ‘computer’ condition, the task was
prefaced by some general questions about whether they enjoyed working
with computers, whether they had one at home, and so on.
The pattern of results in terms of recall scores is shown in Figure 6.2.
As can be seen, the girls did better when the task was presented on paper
than when it was presented on computer, while the reverse was true of the
boys. This resulted in a statistically signiWcant interaction between gender
and condition eVects in an analysis of variance. A simple main eVect
analysis showed the gender diVerence in the ‘on paper’ condition to be
statistically signiWcant.
A similar pattern of results was shown in terms of a diVerent test of
memory, which involved recognition. This was done after the recall task,
and involved pupils being presented with a series of twenty paired images.
Members of each pair were identical except in orientation. Children had
to decide whether the image represented a fragment of the original Wgure
or not, and if so, which image was correctly oriented.
For children in the ‘on paper’ condition, the paired images were
presented as a booklet, with a pair of images on each page. For children
Figure 6.3 The mean recognition score for the girls and boys for the
two modes of task presentation
in the ‘computer’ condition the images were presented on the computer

screen and the children used speciWed keys on the keyboard to make
their responses. Figure 6.3 shows the results; there was a signiWcant
interaction between gender and condition. Girls tended to show better
recognition scores when the task was presented on paper, but boys
scored signiWcantly better than girls when the task was presented on
computer.
Direct comparisons of the same task undertaken on and oV computer
are surprisingly few and far between, so it may be appropriate here to
mention brieXy an unpublished study (Humphreys, 1995) which looked
at the performance of boys and girls on physical and computer screen
versions of the Tower of Hanoi problem described in Chapter 2. Eighty
children, aged ten and eleven, working in coactive pairs (i.e. without
interaction) attempted the Tower of Hanoi task either as a wooden puzzle
or on the screen. While boys and girls took approximately the same time
to complete the physical version of the task, boys were signiWcantly
quicker on the screen version than girls were, resulting in a statistically
signiWcant interaction. However, the same pattern was not shown for the
numbers of moves needed to complete the task; here there was no

signiWcant eVect of gender.
Game versus test context

As well as these studies of how far computer presentation of a task
inXuenced responses diVerentially by gender, we designed a study to
address the question of how far the context in which a computer task is
presented can aVect girls’ and boys’ on-task performance. In an experi-
mental study involving sixty children aged ten and eleven (thirty boys and
thirty girls), we examined the eVects of diVerential contextualisation on
girls’ and boys’ performance on a computer-based perceptual-motor
skills task (Littleton, Ashman, Light, Artis, Roberts and Oosterwegel,
1999).
The task was a specially designed computer version of a physical device
in which the participant attempts to guide a wire ring around a bent wire
frame without making contact with the wire. The screen version consisted
of a double cursor framing an irregular line, which moved across the
screen at a regular speed. Figure 6.4 shows the starting set up as it
appeared to the child. Note that the window labels could be changed so as
to display a stereotypically masculine label (technician’s skills test), a
stereotypically feminine label (beautician’s skills test) or a game label
(electric eel game) as appropriate.
The line moved from right to left across the screen. The objective was
to use the mouse in order to raise and lower the cursor in such a way that
the line would pass smoothly through the gap, without hitting the cursor
‘ring’. If there was contact, a beep sounded and the line stopped until the
cursor was moved so as to free up the line again.
Each child was seen individually in a quiet side room in the school. The
task, with one of the possible labels, was loaded on the screen before the
child came in. After the child had been seated in front of the computer,
the experimenter sat to one side of the computer and gave a scripted
introduction. For example, in the case of the feminine context the in-
structions were as follows: ‘I’m interested to see how children of your age
get along with some computer programs. I’m trying out a few diVerent
things today. The one I’d like you to do is designed as a test to see how
good people will be as beauticians.’ The ending of this introduction was
amended according to the experimental condition. In the case of the
masculine context the children were told that the program was ‘designed
as a test to see how good people will be as technicians’. In the case of the
game context, the children were told that the program was ‘designed as a
computer game and is called electric eel.’
Figure 6.4 The perceptual-motor skills task
The experimenter then ran a ten second demonstration during which

she controlled the mouse and described the task. The children then ran
the demonstration for themselves as a practice session. Just as each child
began to embark upon the session from which their performance data
were derived, the experimenter reinforced the context by saying ‘OK,
now it’s your turn to do the beautician’s skills test /technician’s skills test/
electric eel game.’ The experimenter then recorded details of the per-
formance measure, the number of hits incurred, and the screen was then
cleared so that the next participant would not see the previous scores.
Analysis of the frequency of ‘hits’ (i.e. occasions on which the cursor
ring was allowed to hit the line) revealed no signiWcant main eVects nor
any evidence of a signiWcant interaction between gender and context.
However, from a closer inspection of means it appeared that heterogene-
ity of performance within the skills test conditions was potentially mask-
ing evidence of gender-related performance diVerences in the game con-
dition. We thus decided to undertake a further series of post hoc
comparisons. These revealed a signiWcant simple main eVect of gender in
the game context with the performance of the girls being signiWcantly
worse (i.e. more hits) than that of the boys. There were no signiWcant
gender eVects for either of the two skills tests.
Figure 6.5 The mean ‘hits’ for the girls and boys in the test and game
conditions
The pattern of results thus suggests that the important distinction is not
between the beautician or technician contexts but rather between ‘game’
and ‘test’ contexts. If the context factor is treated as having only these two
levels, the resulting two-factor analysis of variance shows not only that
boys do rather better than girls overall, but more interestingly a signiWcant
interaction between gender and context (see Figure 6.5).
Further analysis reveals that where the task is introduced as a test there
is no trace of a performance diVerence between the girls and the boys. By
contrast, when exactly the same task is introduced as a game the boys’ and
girls’ performance diverges to a point where there is a signiWcant gender
diVerence in performance favouring the boys. Moreover, the girls per-
formed signiWcantly better when the task was described as a test than
when it was described as a game.
The data suggest that the gendered speciWcation of the ‘test’ contex-
tualisation (as beautician or technician) had no eVect on the performance
of girls and boys. More generally the boys’ performance appears to be
very little aVected by the context of the task presentation. This is not a
ceiling eVect, as there was considerable room for improvement. The girls’
performance, on the other hand, was signiWcantly aVected by the particu-

lar context encountered, deteriorating dramatically in the ‘game’ context.
In attempting to address the reasons for this it is perhaps worth noting
that an answer cannot easily be found within the mainstream cognitive
psychology literature. To the archetypal cognitive psychologist, a task is a
task is a task. It possesses clear demand characteristics. It can be deWned
essentially from the outside. Its meaning is determinate. The act of
diVerentially contextualising a task should, from such a point of view,
make no diVerence to a participant’s performance on that task. Yet here
there is compelling evidence that it can matter considerably. Participants’
‘readings’ of a task can, in certain circumstances, impact profoundly on
their performance on-task.
In terms of explaining our Wndings, we can oVer four possible accounts.
All involve the recognition that the interaction between the experimenter,
the task and the child is of paramount importance. An experimenter does
not simply ‘deliver’ a set of instructions to a participant. In introducing a
task to a participant the experimenter is creating a crucial frame of
reference, constructing a context within which the child is enjoined to
work. The participant attempts to build an understanding of what is
expected and of the intentions and motives of the experimenter, through
a process of interpretation and re-interpretation of the experimenter’s
action and interaction.
As noted above, the interaction between experimenter and participant
in an experimental setting is a subtle exercise in ‘sense making’, which will
be shaped in part by the wider cultural context within which the interac-
tion is located. In this particular study the experiment is being conducted
within a school setting. Conducting research studies with children in
schools undoubtedly aVords many distinct advantages, for example, rela-
tive ease of access to large numbers of participants. But such an arrange-
ment may have considerable implications for the data which result from
such studies. Responses may be determined by a complex contextual
system, which extends far beyond the immediate interaction between
experimenter and participant, and which is inseparable from the way that
education and educational activity is deWned in our culture.
We know from the wider educational literature that, as a group, girls
tend to be interested in, and sensitive to, the social context of schooling.
They are highly skilled at ‘making sense of school’ and are more moti-
vated than boys are to understand what it means to be a pupil and what it
means to ‘do’ school tasks (Barber, 1994; Davies and Brember, 1995). A
Wrst line of explanation of girls’ performance in the game context might
thus reXect their uncertainty regarding how to read the intentions and
motives of an unfamiliar adult who, when working in a classroom setting
during lesson time, instructs them to play a computer game. What is the
purpose of such activity? The playing of computer games is not part and
parcel of recognised daily classroom activity. Thus there is a mismatch
between the girls’ understanding of what constitutes an appropriate
school task and the task they were asked to perform. We might thus
speculate that the girls do not know how to make sense of this interaction
with the experimenter, that they are unsure as to the purpose and mean-
ing of the activity they have been asked to undertake, since computer
gaming has nothing to do with the business of schooling.
Seen in these terms, then, the girls’ substantially better performance in
the ‘test’ conditions would reXect the fact that the intentions and motives
of the experimenter and the purpose of the activity is less ambiguous and
more appropriate to a school context. Although the unfamiliar adult is
asking them to perform a computerised vocational skills test, which they
would not be familiar with, a test is an activity wholly appropriate to the
classroom context. Tests are an integral feature of the school experience.
Moreover, adults who are not members of teaching staV (for example,
educational psychologists) can often be seen in school administering tests
to children.
This account rests on the assumption that girls tend to prefer ‘purpose-
ful’ activity and that they demonstrate sensitivity to context in a way that
boys tend not to. This is an assumption which has some grounding in the
wider educational literature. For example, data from the Assessment of
Performance Unit in Science reveal that within the domain of science,
girls take into account the circumstances within which a task is set,
whereas boys, as a group, consider the task in isolation from its context
(Murphy, 1993).
A second possible explanation, also predicated on the notion of girls’
sensitivity to context, concerns the cultural signiWcations of the context
‘game’. There is a vast amount of literature pointing to the signiWcant
association between gender and frequency of electronic game playing
(GriYths, 1991). Males tend to have considerably more computer-game
experience than females, both in childhood (Subrahmanyam and Green-
Weld, 1994) and adulthood (GreenWeld, Brannon and Lohr, 1994). They
have ‘learned how to learn’ video games and have played and practised on
them extensively and enthusiastically. Girls tend not to be enthusiastic
game players. For example, when interviewing children about their
thoughts on video games Stutz found that the girls were less keen than
boys on gaming because they saw the computer world as being biased
towards the male perspective and they were therefore discouraged from
taking an interest in them (Stutz, 1996). It is therefore possible that the
contextualising of the activity as a ‘game’ eVectively categorises the task as
an activity somehow more appropriate for boys. Disengagement from the

task then leads to a relative decrease in the girls’ performance. A variant
on this might posit that the description of the task as a game places the
girls in a position of relative inexperience and it is an anticipation of
failure which becomes a self-fulWlling prophecy.
The further alternative is simply that the girls view the activity of
gaming as a frivolous, unimportant activity and that their performance on
the task is of little consequence. Aspects of identity are not bound up with
being a computer gaming expert. The context of a ‘test’, on the other
hand, may signal the need to try to perform well. Getting good marks on
tests in school is bound up with their sense of pupil identity. The lower
performance of the girls in the game condition may thus simply reXect the
fact that their reading of the task was one where outcome was not
particularly important to them. Less good performance is not an indicator
of failure, but rather a reXection of a lack of concern.
It would be hard to demonstrate experimentally which of these expla-
nations could best account for the girls’ response to the categorisation of
the task as a game. Careful post-task interviewing of the female partici-
pants could have a role to play here. Unfortunately, such data are not
available to us in this study. It may, however, be possible for further
studies to distinguish at least the Wrst of the explanations ventured above
from the others. The Wrst explanation rests on the notion that encounter-
ing a game in a school context is particularly problematic. If this is the case,
then we should not expect this pattern to be replicated if the same study
were to be undertaken in non-school settings.
The present Wndings resonate with those of other researchers interested
in children’s interactions with computer technology. For example, Mal-
one (1981) studied the software preferences of school children and dis-
covered that males had a preference for games or toys which provide
interactive opportunities simply for their own sake. Girls on the other
hand preferred purposeful software which might be classiWed as tools.
Findings such as those reported by Malone and those presented here are
of some signiWcance just at the moment, because of the growth of the
‘edutainment industry’ aimed at both home and school markets. Games
are supposed to be synonymous with motivating children to learn and
making learning fun. Currently, we hear a good deal about the potential
of software with game-like qualities to promote learning (GriYths, 1996)
and computer game formats are increasingly being used as vehicles for
educational tasks. The Wndings reported here suggest that, if we are
concerned about fostering girls’ enthusiasm for computers, then the
presentation of a task as a game may be a hindrance to, rather than a
support for, learning and engagement.
Overview
The results of these various studies show that, quite independently of the
intrinsic diYculty of the tasks concerned, presenting a task on a computer
can make a diVerence to performance. Not only that, but computer
presentation has a diVerential impact on the performance of girls and boys.
As the Wrst result in this section showed, this diVerential eVect is evident
even when the children do not need actually to do anything with the
computer at all. We have also seen that variations in the contextualisation
of the task also makes a diVerence as a function of gender. Thus we are
dealing here with eVects which owe more to individuals’ self-perceptions
and attributions than to their abilities – if indeed we can still meaningfully
separate these. The clear message emerging from this work is that there is
a fundamental interrelationship between the cognitive and the social; the
social and emotional dimensions of learning are not analytically separate
or distinct from the activity of learning. A learning situation is an inex-
tricably social situation. The recognition of this poses considerable
methodological and theoretical challenges. How best to study and con-
ceptualise the learner and the activity of learning? Does it make sense to,
as we have done in places, appeal to the notion ‘ability’ or refer to ‘high
and low achievers’ when we have seen that the children’s on-task per-
formance is contextually determined? In the Wnal chapter we attempt to
address some of these issues.
7 Interaction and learning: rethinking
the issues
Talking and learning

The central thesis of this book is that cognitive development and learning
are fundamentally social processes. Whilst it is undoubtedly important to
consider the role of more experienced cultural actors in promoting devel-
opment and learning, our own concern has been mainly with the contri-
bution of interaction between peers.
Our earliest investigations of children’s collaborative learning and
problem solving addressed the issues of whether and when working
collaboratively in pairs would prove more eVective than working alone.
As Chapters 2 and 4 make clear, we soon amassed compelling evidence
that in certain circumstances two heads were indeed better than one. This
research, which focused primarily on the outcomes and products of
collaborative work, was later complemented by a series of studies design-
ed to shed light on the processes of collaborative learning. At Wrst our
analyses of the talk and joint activity of children were informed by
neo-Piagetian conceptions of ‘socio-cognitive conXict’. Our primary re-
search question was whether individual progress in understanding could
be promoted through exposure to the conXicting ideas of a peer in the
context of collaborative problem solving. Whilst the notion of socio-
cognitive conXict aVorded some explanatory power, it was also apparent
that more often than not collaborative gains seemed to have very little to
do with ‘decentring through conXict’ in Doise’s sense (Doise, 1990).
Rather, progress appeared to be associated with socially mediated pro-
cesses of conXict resolution. The associated concepts of argumentation
and negotiation led us to adopt an inherently social model of productive
interaction inspired by ‘social constructivist’ approaches emphasising
processes of joint construction of understanding.
Underpinned by conceptual and methodological tools derived from
neo-Piagetian and neo-Vygotskian traditions of research, our work has
focused primarily on examining the relationship between particular forms
of talk and subsequent individual cognition. Our research strategy has
91
been to use correlational techniques to determine whether there is evi-

dence of an association between particular features of the learners’ talk
(identiWed using categorical coding schemes) and individual learning
gains. In common with many others working in this Weld, we have treated
learning in terms of conceptual accomplishments at an individual level,
demonstrable through various forms of post-test. However, in recent
years a more radical perspective has been emerging in the research
literature on collaboration and learning, a perspective in which social
interaction is seen not just as a stimulus to individual thinking but as, in
eVect, ‘a social mode of thinking’ (Mercer and Wegerif, 1998).
Such a conceptualisation implies that ‘talk and social interaction are not
just the means by which people learn to think, but also how they engage in
thinking . . . discourse is cognition is discourse . . . One is unimaginable
without the other’ (Resnick, Pontecorvo and Säljö, 1997, p.2). As Resnick
and colleagues recognise, the notion that concepts and ideas are con-
stituted in interactive discourse represents a challenge to traditional
accounts of the nature of knowledge. It implies that ‘ways of thinking are
embedded in ways of using language’ (Wegerif and Mercer, 1997, p.51).
This view of cognition challenges our traditional conceptions of devel-
opment and learning. It invites us to reject a conception of the develop-
mental process as the creation of an autonomous thinker who, through a
reconstructive process of internalisation, has acquired the resources and
tools of a culture (Packer, 1993). Gone is the notion of learning as mental
reorganisation. In its place is a view of learning as intersubjective and
dialogical. Learning is about participation and engagement in shared
cultural practices. It involves the acquisition of ‘both the organising
conceptual theories and the patterns of discourse used by particular
reasoning communities’ (Resnick et al., 1997, p.4).
Were we to follow the argument regarding the socio-construction of
knowledge to its logical conclusion then we would be left with a view of
cognition being not only coconstructed but also distributed between
people. In strong form, the argument would be that knowledge inheres in
relationships and only exists whilst those relationships remain intact. If
these relationships become void, the conditions of distribution no longer
apply, and it is not clear what, if anything, the individual is left ‘knowing’.
At limit, there is a ‘person-in-situation, but no personal history’ (Snow,
1994, p.7). As Resnick et al. remark: ‘the situativity point of view seems to
make the individual disappear or at least to exist only when particular
others are about. It is as if the individual were recreated de novo in each
new situation’ (1997, p.5).
Whilst we agree with the proposition that ways of thinking are embed-
ded in the use of language in social context and that learning is intersub-
Interaction and learning 93
jective and dialogical, we reject a ‘strong’ version of this approach on the

grounds that it risks such ontological nihilism (Light, Sheldon and Wood-
head, 1991). What is needed is an approach which respects the funda-
mentally social nature of cognition and yet at the same time does not
unduly diminish the status of the individual. The ‘problem’ of situating
the individual in relation to such an account of cognition is acknowledged
by Resnick et al., who appeal to a notion of a ‘history of experience’:
. . . think of individuals as passing through a series of temporally linked situations.
In each situation, the individual brings to a new interaction, with a particular set
of other people and artefacts, a brain tuned to respond easily and automatically to
particular aVordances and constraints. When tunings and aVordances are suY-
ciently matched, the individual can enter into the particular interactive situation,
both responding to others and shaping their responses so that a mutually con-
stituted set of cognitive actions is possible. (Resnick et al. 1997, p.5).
There is some merit in invoking the concept of a ‘history of experience’,

but greater emphasis needs to be placed on the processes of individual
construction. Contemporary situated and discursive approaches to learn-
ing take the view that the small group or dyad is ‘the crucible of social
meaning’ while culture is ‘the main resource for frameworks of meaning’
(Haste, 1993, p. 186). Accepting this, it also remains the case that
children are agents in their own social construction. Cole (1996) articu-
lates this nicely in observing that mind emerges in joint activity and that
there is a pivotal role for interpretation. This claim hints at the sense in
which there is a dialectical relationship between the transformative
changes on the intra-mental and inter-mental planes. To this extent the
individual can be conceived of ‘as group’ and the structure of agency
understood through an analysis of group functioning (Wertsch, Tulviste
and Hagstrom, 1993). But this is not at all the same as saying that agency
is solely an interpersonal matter. Agency is an individual/cultural dialec-
tic, with development and learning emerging in activities situated in
particular institutional or cultural settings. Children, for example, may
appropriate the particular forms of language, or ‘discourses’, characteris-
tic of formal schooling. In doing so, they are also appropriating culturally
constructed ways of thinking. But appropriation is an action, undertaken
by an agent, and that agency is not negated by the appropriation.
The cultural context of collaboration

By invoking a notion of agency as interpersonal and socio-cultural we
acknowledge the need to understand processes of joint knowledge con-
struction, through careful analyses of discourse which focus on the
continual, subtle, evolutionary process of negotiation and renegotiation

of meaning. We also acknowledge the need to understand the particular
historical, institutional and cultural context of collaborative learning.
Discourse in any socially deWned setting is nested within the wider socio-
cultural context (Valsiner, 1997). As Bruner puts it, learning and think-
ing: ‘are always situated in a cultural setting and always dependent upon
the utilisation of cultural resources’ (1996, p.4). Learning, in other
words, is culturally based, not just culturally inXuenced, and the groups
of children we study are not undertaking their joint work ‘in a vacuum’.
When focusing our research eVorts on microgenetic analyses of sessions
of talk and joint activity we must be careful not to neglect the other
important facets of the picture. We need to recognise that understanding
contexts for collaborative learning involves more than understanding how
the immediate joint activity is resourced. Learners’ interactions are
framed by, and therefore can only ever be fully understood within, the
context of particular institutional structures and settings. As Crook and
Light (in press) put it, our traditions of organised education have evolved
various forms of community structure, and it is important that the impact
of such structures is understood. Observable interactions are likely to
have unobservable determinants in the histories of individuals, groups
and institutions:
The social world inXuences the individual not only through the agency of Xesh-
and-blood people who converse, communicate, model or persuade, but through
the social practices and objects unseen people have built up in the world around
that individual. (White, 1996, xiv)
DiVerent institutional contexts aVord diVerent opportunities for, and

constraints upon, interaction. The educational practices we see today are
the result of a long period of historical development, and the activities of
today’s students are largely circumscribed by existing practices and estab-
lished materials (Crook and Light, in press; Light and Light, 1998). The
implication, then, is that collaborative interactions need to be understood
within their cultural niche, with reference to the broader social and
historical context within which they are positioned. As Mercer highlights:
Children’s interpretations of experience, the meanings they attach to their learn-
ing – will, in part, be determined by their involvement with schools and other
institutions of their society. Any educationally oriented study of learning must
recognise that schools have their own body of cultural knowledge, and their own
ways of communicating and legitimising knowledge. (1992, p.31)
The argument that we need to understand the wider social context of

collaborative working does not only apply to those who do observational
research on collaborative work within classroom settings. The need to
contextualise the study of collaborative learning applies equally to the

experimental tradition of research. In particular, whilst they were not an
integral part of on-going classroom activity, the experimental studies of
collaborative activity described in the previous chapters were undertaken
in school contexts and as such we need to consider the relations between
the experiments and ‘indigenously organised activities’ (Cole, 1996, p.
250).
The phrase ‘locating the experiment’ was Wrst coined by Scribner
(1975) and it is a phrase which highlights that experimental tasks are not
decontextualised occurrences. It also draws attention to the fact that joint
activity undertaken in experimental sessions is shaped by broader dimen-
sions of the context. Responses to experimental tasks undertaken in
educational settings are inXuenced by contextual systems which extend
far beyond the immediate interaction between experimenter and partici-
pants. By studying interactions conducted in school settings we are
studying interactions in a particular ‘niche’. The children we study en-
gage in heavily contextualised discourse. Their interactions are in-
Xuenced by a complex contextual system which is inseparable from how
education is deWned in our culture (Mercer and Fisher, 1992). It is not
the case that the pupils talk explicitly about social structures. Rather,
‘institutional settings are made relevant by participants as part of their
own communicative practices’ (Edwards, 1997, p. 33). Context does not
determine what is said, rather the children construct through talk the
context they need to support that talk (Edwards and Furlong, 1978).
Educational processes thus function as a means of acquiring and sharing
cultural knowledge, knowledge which contextualises subsequent activ-
ities and problems (Mercer, 1992). Social structures are thereby con-
stituted by ‘human agency and yet at the same time are the very medium
of human agency’; each is transformed by the other (Giddens, 1975,
p. 121).
In highlighting the need to understand the institutional context of the
activity we observe, we are also highlighting the need to push back
traditional disciplinary boundaries. As Cole points out:
When we reach this level of analysis we arrive at the borders of what is ordinarily
counted as psychology. As a rule, the social-institutional context of action is
treated as a (largely unanalysed) dichotomised independent variable . . . or left to
sociologists (1996, p. 340).
Pushing at these boundaries also necessitates methodological plural-

ism. There is no one ‘right’ way to study the institutional contexts of
activity. There remains a place for experimental work such as ours which
oVers a very particular window through which to study context and
cognition. Whilst many would undoubtedly criticise our approach on the

grounds that ‘real’ psychological processes are only to be found in natu-
ralistic settings, we would argue that there is no one setting in which ‘true’
psychological functioning and processes are revealed. Contexts for study-
ing cognition, no matter how ‘ecologically valid’, are always constructed.
Though we have for the most part been presenting the results of experi-
ments, we recognise that there is value in other very diVerent approaches
such as, for example, classroom ethnography. Moreover, researchers
interested in educational discourse have much to learn from studying
discourse in other institutional settings (Edwards, 1997). The essential
nature of collaborative classroom talk will never be understood by focus-
ing our analytic energies solely on the study of classrooms. Only through
linking and contrasting collaborative educational discourse with that
observed in other cultural settings will we gain insight into what are the
particular or deWning features of such talk (Edwards, 1997).
Artefacts and aVordances

So far we have talked about the important sense in which learning is
constituted in institutionalised discourse. However, to focus solely on
discourse processes neglects a further important sense in which cognition
is constituted – namely, through our use of tools and artefacts. As Good-
man (1976) maintains, tools and artefacts are ‘world making’. The choice
of which artefact to use aVects the structure of our work activity (Scribner
and Cole, 1981), thereby fundamentally transforming the cognitive and
communicative requirements of our actions (Säljö, 1995, p.90). Artefacts
are both ideal and material. White (1996, p. xii) describes them as
‘changeling objects’ that are at one and the same time things and ‘reposi-
tories of prior human thought’.
Säljö (1995) stresses that if we are to understand cognition and practi-
cal action, we must recognise that people and artefacts operate in a system
that cannot be divided. There is cofunctionality to the extent that the
mediational means form part of actions in situated practices. This in turn
has considerable implications for the ways in which we study learning.
For example, in the case of our work, we have frequently turned to the
study of children working together at a computer. The computer is an
artefact which is not only capable of supporting collaborative endeavour,
but has the potential to transform the way in which collaborative activity
is organised. This requires us to conceptualise the role of the computer in
framing and mediating joint activity.
Understanding the mediating role of the computer is a recurrent theme
in much recent work. The research literature has oVered a number of
striking illustrations of the ways in which computer software structures

and re-organises the social processes of problem solving and teaching and
learning (Järvelä, 1995, Golay-Schilter, Perret, Perret-Clermont and De
Guglielmo, 1998). Our own work has also highlighted the role computer
equipment and interface devices play in mediating joint activity. For
example, in the case of the studies described in Chapter 2, the use of a
keyboard dual input device was vital in ensuring that the children engaged
with one another as well as with the task in hand.
It is important to remember, however, that we never experience arte-
facts in isolation, but only in connection with a contextual whole. An
object ‘. . . is always a special part, phase or aspect of an environing
experienced world’ (Dewey, 1938, p.67). So a child’s reaction to and
performance on a computer task may be crucially determined by the
context of activity within which the task is encountered. The work dis-
cussed in Chapters 5 and 6, where we investigated the eVects of diVeren-
tial contextualisation on individual girls’ and boys’ performance on com-
puter-based problem-solving and perceptual-motor skills tasks, bears
striking testimony to this. Even with highly constrained tasks, seemingly
small changes to the contextualisation of the task can be highly signiW-
cant. Moreover such changes can aVect not just the absolute diYculty of a
task, but also its relative diYculty for diVerent groups of children.
Whilst these studies represent powerful demonstrations of the interre-
lationship between learner and context, the Wndings are often hard to
interpret. Consider for example the case of the work on the perceptual-
motor skills task described in Chapter 6. We rehearsed there numerous
possible explanations for these Wndings. Our inability to settle on a
preferred explanation highlights the diYculties associated with interpret-
ing apparent ‘failure’ under such circumstances. We are confronted with
both the necessity and the diYculty of adopting the child’s point of view
regarding the task in hand, and the importance of attempting to under-
stand a participant’s goals and frame of reference, as opposed to working
with our own assumptions concerning what these are or should be.
As researchers, we are so often bounded by our own subjectivity and
our motivating research questions that we fail to ‘decentre’ suYciently to
recognise or adopt other interpretative frameworks and points of view. As
Grossen and Perret-Clermont’s (1997) analyses of researcher–child in-
teraction in experimental contexts illustrates, our own limited interpreta-
tions of what counts as a good, correct or appropriate response in such
settings mean that we often fail to recognise the creative responses
children make in such settings. All too often we treat children as objects of
concern, rather than people with concerns (Prout, 1998). As psychologists
we may be locked into a process of ‘negative scholarly rationalism’ (Säljö,
1997). We readily attribute children’s behaviour to psychological incom-

petence, rather than seeking to develop our understanding of the activity
of the child in context. Rather than theorising incompetence, we need to
understand how the situations in which children are working and the
meanings they ascribe to tasks support or constrain their activity and
performance. Thus:
So long as psychologists continue to work within a conventional ‘developmental’
framework, they will be open to the charge that they emphasise relative incompe-
tence, immaturity and dependency in ways that implicitly diminish children’s
status. . . . Notions of ‘competence’ [should be] viewed as problematic, informed
by cultural beliefs and negotiated by participants in particular social contexts.
(Woodhead, Faulkner and Littleton, 1999)
The implication of this line of argument is that we need to do more than

simply be open to diVerent interpretations of our research Wndings. We
need to recognise that a child’s performance in experimental test situ-
ations is always situated. It is not simply the case that some experimental
tests are somehow ‘fairer’ than others and are therefore better at ‘reveal-
ing’ children’s competence. Children’s performance in a test situation
represents the ‘tuning of particular persons to the particular demands and
opportunities of a situation, and thus resides in the combination of
person-in-situation, not ‘‘in the mind’’ alone’ (Snow, 1994, p. 31).
If we accept the idea of ‘person-in-situation’ as the appropriate unit of
analysis, then the concept of ‘aVordance’ (Gibson, 1977) is relevant. Put
simply, the term ‘aVordance’ refers to what a situation oVers a person. As
Snow makes clear, the term implies:
a complementarity of person and situation, as in an ecological niche . . . AVordan-
ces reXect the invitation, demand, or opportunity structure of a situation for those
persons who are tuned or prepared to receive them. Particular aVordances invite
particular actions. The potential actions of which a person is capable are called
eVectivities . . . Abilities are thus unique coalitions of aVordances and eVectivities
in particular person-treatment systems. (Snow, 1994, pp. 28–29)
The conception of abilities as linked to aVordances emphasises the

importance of the ‘attunement’ between person and situation. Moreover,
if the person-in-situation becomes our analytic unit then we need to
understand more about how people come to be ‘tuned’ and how and why
they perceive particular aVordances in a situation. Objects may in some
sense signal their aVordances, but attunement is not just about the child
making sense of the world of objects. Situations are psychological spaces
as well as physical spaces requiring what Rommetveit (1990) calls an
attunement to the attunement of others. Children’s responses and reac-
tions to the tasks psychologists give them are crucially dependent upon
the negotiation of the experimental or didactic contract (Grossen and

Pochon, 1997), and the coconstruction of meaning between experimen-
ter and child. As Grossen and Pochon note, test situations are ‘communi-
cation situations that are culturally rooted and whose meanings have to be
constructed intersubjectively during the interaction’ (1997, p. 269).
Learning, identity and the social basis of development

Approaching the study of child development and learning in social and
cultural contexts involves abandoning ‘the conventional demarcation
between cognitive, social and emotional development’ (Woodhead, Faul-
kner and Littleton, 1998, p. 1). It no longer makes sense to say that we
need to study the social and emotional dimensions of learning as if such
dimensions were analytically separate and distinct from the activity of
learning itself. One of the key features of a social/emotional approach to
cognition is the recognition that the processes of thinking and identity
construction go hand in hand:
to think or to reason well in a situation is, by deWnition, to take on the forms as
well as the substance of a community of reasoners and thus to join that commu-
nity. Much of discourse, and thus [much] of cognition serves to situate an
individual with respect to others, to establish a social role or identity (Resnick,
Pontecorvo and Säljö 1997, p. 9)
The importance of identity, and the ways in which children position

themselves with respect to a task or their peers, was demonstrated in
Chapters 5 and 6. Here we have seen how attributions and perceptions of
expertise are vital elements in cognitive activity. From a very early age
children are engaged in the construction of their identities as pupils. They
construct and participate in discourses about ability and eVort (Bird,
1994) and are motivated to understand what it means to be a learner and
what it means to succeed at educational tasks. The social climate of
comparison, competition, success, failure and issues of relative status in
the classroom rapidly becomes established within the early years of
schooling (Crocker and Cheesman, 1988) and remains a powerful inter-
personal dynamic throughout children’s educational careers.
This view of development and learning being sketched here has far
reaching implications for the way we as developmental psychologists ‘do’
our science. Developmental psychology is a partial science; a science of
invention, not of discovery. Simple faith in the objectivity of scientiWc
testing will not suYce. What is needed is a detailed understanding of
issues of subjectivity, intersubjectivity and social construction. We are not
engaged in the activity of discovering the ‘real’ capabilities of the child.
Rather we are involved in the business of constructing particular accounts

and representations of the child and notions of competence (Woodhead,
Faulkner and Littleton, 1999). The representations we create are histori-
cally located, culturally determined and value-laden, as are the very issues
we choose to investigate. As Burman puts it: ‘the developmental psychol-
ogy we know is tied to the culture which produced it’ (Burman, 1999).
Our own interest in children’s use of information and communications
technology, for example, has clearly been shaped by unarticulated cul-
tural preconceptions about the contexts, goals and processes of develop-
ment in Western society at the end of the twentieth century (Woodhead,
Faulkner and Littleton, 1999).
Developmental psychologists do not exist in isolation from the social
contexts they study. We too are ‘situated’ in particular institutional,
cultural and historical contexts and it is in such contexts that we actively
create our subject. Developmental theories and research shape the envi-
ronments in which children develop and learn, and to this extent psychol-
ogists participate in the construction of contemporary reality (Rose,
1990; Woodhead, 1999). The psychologist, in this sense, is in no diVerent
a position to any other learner, engaged in a continuing intersubjective
negotiation of meaning within the ever-present constraints of the mat-
erial world.
Our hope is that the approach we have arrived at here will illuminate
some educational as well as some psychological issues. In an era of
‘standardised assessment tests’ and the like, designed to drive up educa-
tional standards by highlighting individual success and failure, there may
be merit in drawing attention to the social and relational bases of such
achievements. We have seen that ‘two heads can be better than one’ and
that ‘wholes can be more than the sum of their parts’. We have presented
evidence that the exercise as well as the acquisition of understanding is
underpinned by complex social processes. Learning, whether in the class-
room or in cyberspace, is a process of engagement with culturally elabor-
ated and socially mediated reality. The social processes which shape such
learning are, on the evidence we have reviewed, powerful in their eVects.
The harnessing of these processes to support children’s learning holds the
key to enhancing the eVectiveness of their education, in the widest sense.
If the studies gathered together in this volume contribute in some small
way to this objective, then they will have served their purpose.
References
Ames, C. (1981) ‘Competitive vs. cooperative reward structures: The inXu-

ence of individual and group performance factors on achievement,
attributions and aVect’, American Educational Research Journal, 18, pp.
273–87.
Azmitia, M. and Montgomery, R. (1993) ‘Friendship, transactive dialogues,
and the development of scientiWc reasoning’, Social Development, 2, pp.
202–21.
Barber, M. (1994) Young people and their attitudes to school (Interim Report),
Keele, University Centre for Successful Schools.
Barbieri, M. S. and Light, P. (1992) ‘Interaction, gender and performance on
a computer based problem solving task’, Learning and Instruction, 2, pp.
199–214.
Barron, A. and Foot, H. (1991) ‘Peer tutoring and tutor training’, Education-
al Research, 33, pp. 174–85.
Bennett, N., Desforges, C., Cockburn, A. and Wilkinson, B. (1984) The
Quality of Pupils’ Learning Experiences, London, Erlbaum.
Berndt, T., Perry, T. and Miller, K. (1988) ‘Friends’ and classmates’ interac-
tions on academic tasks’, Journal of Educational Psychology, 80, pp.
506–13.
Beynon, J. (1993) ‘Computers, dominant boys and invisible girls: Or ‘‘Han-
nah, it’s not a toaster, it’s a computer!’’’ in Beynon, J. and MacKay, H.
(eds.) Computers into Classrooms: More Questions than Answers, London,
Falmer Press.
Bird, L. (1994) ‘Creating the capable body: Discourses about ability and
eVort in primary and secondary school studies’ in Mayall, B. (ed.)
Children’s Childhoods: Observed and Experienced, London, Falmer Press.
Blaye, A. and Light, P. (1995) ‘The inXuence of peer interaction on planning
and information handling strategies’ in O’Malley, C. (ed.) Computer
Supported Collaborative Learning, Berlin, Springer Verlag.
Blaye, A., Light, P., Joiner, R. and Sheldon, S. (1991) ‘Collaboration as a
facilitator of planning and problem solving’, British Journal of Develop-
mental Psychology, 9, pp. 471–83.
Bliss, J., Askew, M. and Macrae, S. (1996) ‘EVective teaching and learning:
ScaVolding revisited’, Oxford Review of Education, 22 (1), pp. 37–61.
Bond, C. and Titus, L. (1983) ‘Social facilitation: A meta-analysis of 241
studies’, Psychological Bulletin, 94, pp. 265–92.
101
102 References
Bone, A. (1983) Girls and Girls-only Schools: A Review of the Evidence, Man-
chester, Equal Opportunities Commission.
Bruner, J. (1985) ‘Vygotsky: A historical and conceptual perspective’ in
Wertsch, J. (ed.) Culture, Communication and Cognition: Vygotskian Per-
spectives, Cambridge, Cambridge University Press.
Bruner, J. (1996) The Culture of Education, Cambridge, MA, Harvard Univer-
sity Press.
Buckley, P. and Smith, P. (1991) ‘Opting out of technology: A study of girls’
GCSE choices’ in Lovegrove, G. and Segal, B. (eds.) Women into Com-
puting, Heidelberg, Springer-Verlag.
Burman, E. (1999) ‘Morality and the goals of development’ in Woodhead,
M., Faulkner, D. and Littleton, K. (eds.) Making Sense of Social Develop-
ment, London, Routledge, 170–80.
Burnham, W. (1905) ‘The group as a stimulus to mental activity’, Science, 31,
pp. 761–76.
Butler, R. and Ruzany, N. (1993) ‘Age and socialisation eVects on the
development of social comparison motives and normative ability assess-
ment in kibbutz and urban children’, Child Development, 64, pp. 532–43.
Butterworth, G. and Light, P. (1982) Social Cognition: Studies in the Develop-
ment of Understanding, Chicago, University of Chicago Press.
Camp, T. (1997) ‘The incredible shrinking pipeline’, Communications of the
Association for Computing Machinery (ACM), 40 (10), pp. 103–11.
Clements, D. (1987) ‘A longitudinal study of the eVects of Logo program-
ming on cognitive abilities and achievement’, Journal of Educational
Computing Research, 3 (1), pp 73–94.
Clements, D. and Nastasi, B. (1992) ‘The role of social interaction in the
development of higher order thinking in Logo environments’ in De
Corte, E., Linn, M., Mandl, H. and VerschaVel, L. (eds.) Computer-
Based Learning Environments and Problem Solving, Berlin, Springer-
Verlag.
Cole, M. (1996) Cultural Psychology: A Once and Future Discipline, Cam-
bridge, MA, The Belknap Press of Harvard University Press.
Cooper, J., Hall, J. and HuV, C. (1990) ‘Situational stress as a consequence
of sex-stereotyped software’, Personality and Social Psychology Bulletin,
16, pp. 419–29.
Costa, E. (1991). ‘The present and future of intelligent tutoring systems’ in
Scanlon, E. and O’Shea, T. (eds.) New Directions in Educational Technol-
ogy, Berlin, Springer-Verlag.
Cottrell, N., Wack, D., Sekerak G. and Rittle, R. (1968) ‘Social facilitation of
dominant responses by the presence of an audience and the mere pres-
ence of others’, Journal of Personality and Social Psychology, 47, pp.
1391–98.
Crocker, T. and Cheesman, R. (1988) ‘The ability of young children to rank
themselves for academic ability’, Educational Studies 14 (1), pp. 105–10.
Crook, C. (1987) ‘Computers in the classroom: DeWning a social context’ in
Rutkowska, J. and Crook, C. (eds.) Computers, Cognition and Develop-
ment, Chichester, Wiley.
References 103
Crook, C. (1994) Computers and the Collaborative Experience of Learning,

London, Routledge.
Crook, C. and Light, P. (in press) ‘Information technology and the culture of
student learning’ in Bliss, J., Light, P. and Säljö, R. (eds.) Learning Sites:
Social and Technological Contexts for Learning, Oxford, Pergamon.
Culley, L. (1988) ‘Girls, boys and computers’, Educational Studies, 14, pp.
3–8.
Culley, L. (1993) ‘Gender equity and computing in secondary schools’ in
Beynon, J. and MacKay, H. (eds.) Computers into Classrooms: More
Questions than Answers, London, Falmer Press.
Cummings, R. (1985) ‘Small group discussions and the microcomputer’,
Journal of Computer Assisted Learning, 1, pp. 149–58.
Dalton, D. (1990) ‘The eVects of cooperative learning strategies on achieve-
ment and attitudes’, Journal of Computer Based Instruction, 17, pp. 8–16.
Davies, J. and Brember, I. (1995) ‘Attitudes to school and the curriculum in
Year 2, Year 4 and Year 6: Changes over four years’, Paper presented at
the European Conference on Educational Research (ECER), Bath, Eng-
land, September.
De Corte, E., VerschaVel, L. and Schrooten, H. (1992) ‘Cognitive eVects of
learning to program in Logo’ in De Corte, E., Linn, M., Mandl, H. and
VerschaVel, L. (eds.) Computer-Based Learning Environments and Problem
Solving, Berlin, Springer-Verlag.
Dewey, J. (1938) Experience and Education, New York, Macmillan.
Doise, W. (1990) ‘The development of individual competencies through
social interaction’ in Foot, H. Morgan, M. and Shute, R. (eds.) Children
Helping Children, Chichester, John Wiley.
Doise, W. and Mugny, G. (1984) The Social Development of the Intellect,
Oxford, Pergamon.
Doise, W., Mugny, G. and Perret-Clermont, A.-N. (1975) ‘Social interac-
tion and the development of cognitive operations’, European Journal of
Social Psychology, 5, pp. 367–83.
Doise, W., Mugny, G. and Perret-Clermont, A.-N. (1976) ‘Social interac-
tion and cognitive development’, European Journal of Social Psychology, 6,
pp. 245–47.
Donaldson, M. (1978) Children’s Minds, London, Fontana.
Durkin, K. (1995) Developmental Social Psychology, Oxford, Blackwell.
Edwards, D. (1997) ‘Towards a discursive psychology of classroom educa-
tion’ in Coll, C. and Edwards, D. (eds.) Teaching, Learning and Classroom
Discourse, Madrid, Fundación Infancia y Aprendizaje.
Edwards, D. and Furlong, V. (1978) The Language of Teaching and Learning,
London, Heinemann.
Edwards, D. and Mercer, N. (1987) Common Knowledge: The Development of
Understanding in the Classroom, London, Methuen.
Festinger, L. (1954) ‘A theory of social comparison processes’, Human
Relations, 7, pp. 117–40.
Fitzpatrick, H. (1996) ‘Peer collaboration and the computer’, Unpublished
PhD thesis, University of Manchester.
104 References
Fitzpatrick, H. and Hardman, M. (1994) ‘Gender and the classroom com-

puter: Do girls lose out?’ in Foot, H., Howe, C., Anderson, A., Tolmie,
A. and Warden, D. (eds.) Group and Interactive Learning, Southampton,
Computational Mechanics Publications.
Flynn, E. (1995) ‘The eVects of own gender and partner gender on children’s
computer-based problem solving in the context of a single sex school’,
Unpublished dissertation, Department of Psychology, University of
Southampton.
Forman, E. and Cazden, C. (1985) ‘Exploring Vygotskian perspectives in
education: The cognitive value of peer interaction’ in Wertsch, J. (ed.)
Culture, Communication and Cognition: Vygotskian Perspectives, Cam-
bridge, Cambridge University Press.
Forman, E., Minick, N. and Stone, A. (1993) Contexts for Learning: Sociocul-
tural Dynamics in Children’s Development, Oxford, Oxford University
Press.
Gabrenya, W., Wang, Y. and Latane, B. (1985) ‘Social loaWng on an optimis-
ing task: Cross cultural diVerences amongst Chinese and Americans’,
Journal of Cross Cultural Psychology, 16, pp. 223–42.
Galton, M. and Williamson, J. (1992) Groupwork in the Primary School,
London, Routledge.
Gardner, J., McEwan, A. and Curry, C. (1985) ‘A sample survey of attitudes
to computer studies’, Computers and Education, 10, pp. 293–98.
Gauvain, M. and RogoV, B. (1989) ‘Collaborative problem solving and
children’s planning skills’, Developmental Psychology, 25, pp. 131–51.
Gergen, K. (1977) ‘The social comparison of self knowledge’ in Mischel, T.
(ed.) The Self: Psychological and Biological Issues, NJ, Rowman and Little-
Weld.
Giacquinta, J., Bauer, J. and Levin, J. (1993) Beyond Technology’s Promise,
Cambridge, Cambridge University Press.
Gibson, J. (1977) ‘The theory of aVordances’ in Shaw, R. and Bransford, J.
(eds.) Perceiving, Acting and Knowing: Toward an Ecological Psychology,
Hillsdale, NJ, Lawrence Erlbaum.
Giddens, A. (1975) New Rules of Sociological Method, New York, Basic Books.
Glachan, M. and Light, P. (1982) ‘Peer interaction and learning: Can two
wrongs make a right?’ in Butterworth, G. and Light, P. (eds.) Social
Cognition: Studies of the Development of Understanding, Brighton, Har-
vester Press.
Glaser, A. (1982) ‘Drive theory of social facilitation: A critical appraisal’,
British Journal of Social Psychology, 21, pp. 265–82.
Golay-Schilter, D., Perret, J.-F., Perret-Clermont, A.-N. and De Guglielmo,
F. (1999) ‘Socio-cognitive interactions in a computerised industrial task:
Are they productive for learning?’ in Littleton, K. and Light, P. (eds.)
Learning with Computers: Analysing Productive Interaction, London, Rout-
ledge.
Goodman, N. (1976) Languages of Art, Indianapolis, IN, Hackett.
Graddol, D. and Swann, J. (1989) Gender Voices, Oxford, Blackwell.
GreenWeld, T. (1995) ‘Sex diVerences in science museum exhibit attraction’,
Journal of Research in Science Teaching, 32, pp. 925–38.
References 105
GreenWeld, P., Brannon, C. and Lohr, D. (1994). ‘Two dimensional repre-

sentation of movement through three dimensional space: The role of
video-game experience’, Journal of Applied Developmental Psychology, 15,
pp. 87–103.
GreenWeld, P. and Lave, J. (1982) ‘Cognitive aspects of informal education’
in Wagner, D. and Stephenson, H. (eds.) Cultural Perspectives on Child
Development, San Francisco, Freeman.
GriYths, M. (1991) ‘Amusement machine playing in childhood and adoles-
cence: A comparative analysis of video games and fruit machines’, Jour-
nal of Adolescence, 14 , pp. 53–73.
GriYths, M. (1996) ‘Computer game playing in children and adolescents: A
review of the literature’ in Gill, T. (ed.) Electronic Children: How Children
are Responding to the Information Revolution, London, National Children’s
Bureau.
Grossen, M. (1988) ‘La construction sociale de l’intersubjectivité entre adult
et enfant en situation de test’, PhD thesis, University of Neuchatel,
Switzerland.
Grossen, M. and Perret-Clermont, A.-N. (1997) ‘Learning: A reproduction
of institutionalised teaching practices or a personal appropriation’, Paper
presented at the Seventh European Conference for Research on Learn-
ing and Instruction, Athens, August.
Grossen, M. and Pochon, L.-C. (1997) ‘Interactional perspectives on the use
of the computer and on the technological development of a new tool:
The case of word processing’ in Resnick, L., Säljö, R., Pontecorvo, C.
and Burge, B. (eds.) Discourse, Tools and Reasoning: Essays on Situated
Cognition, Berlin and New York, SpringerVerlag.
Hartley, J. (1968) ‘Some factors aVecting student performance in pro-
grammed learning’, Programmed Learning and Educational Technology, 5,
pp. 206–18.
Hartley, J. and Hogarth, F. (1971) ‘Programmed learning in pairs’, Educa-
tional Research, 13, pp. 130–34.
Hartwick, J. and Nagao, D. (1990) ‘Social facilitation eVects in recognition
memory’, British Journal of Social Psychology, 29, pp. 193–210.
Haste, H. (1993) ‘Morality, self and sociohistorical context: The role of lay
social theory’ in Noam, G. and Wren, T. (eds.) The Moral Self, Cam-
bridge, MA, MIT Press.
Hattie, J. and Fitzgerald. D. (1988) ‘Sex diVerences in attitudes, achievement
and use of computers’, Australian Journal of Education, 31, pp. 3–26.
Healy, L., Pozzi, S. and Hoyles, C. (1995) ‘Making sense of groups,
computers and mathematics’, Cognition and Instruction, 13, pp.
505–23.
Hogan, D. and Tudge, J. (in press) ‘Implications of Vygotskian theory for
learning’ in O’Donnel, A. and King, A. (eds.) Cognitive Perspectives on
Peer Learning, Matwah, NJ, Lawrence Erlbaum Associates.
Howe, C. and Tolmie, A. (1999) ‘Productive interaction in the context of
computer supported collaborative learning in science’ in Littleton, K.
and Light, P. (eds.) Learning with Computers: Analysing Productive Inter-
action, London, Routledge.
106 References
Howe, C., Tolmie, A., Anderson, A. and Mackenzie, M. (1992) ‘Conceptual

knowledge in physics: The role of group interaction in computer sup-
ported teaching’, Learning and Instruction, 2, pp. 161–84.
Hoyles, C. (ed.) (1988) Girls and Computers, London, Institute of Education,
Bedford Way Papers, 34.
Hoyles, C. and Sutherland, R. (1989) Logo Mathematics in the Classroom,
London, Routledge.
Hoyles, C., Sutherland, R. and Healy, L. (1991) ‘Children talking in com-
puter environments’ in Durkin, K. and Shire, B. (eds.) Language in
Mathematical Education: Research and Practice, Milton Keynes, Open
University Press.
Hughes, M. (1990) ‘Children’s computation’ in Grieve, R. and Hughes, M.
(eds.) Understanding Children, Oxford, Blackwell.
Hughes, M., Brackenridge, A., Bibby, A. and Greenhough, P. (1988) ‘Girls,
boys and turtles: Gender eVects in young children’s learning with Logo’
in Hoyles, C. (ed.) Girls and Computers, London, Institute of Education,
Bedford Way Papers, 34.
Hughes, M., Brackenridge, A. and MacLeod, H. (1987) ‘Children’s ideas
about computers’ in Rutkowska, J. and Crook, C. (eds.) Computers,
Cognition and Development, Chichester: John Wiley.
Hughes, M., Greenhough, P. and Laing, K. (1992) ‘Gender eVects in young
children learning Logo’. Final report to the NuYeld Foundation, Exeter:
School of Education, University of Exeter.
Huguet, P., Charbonnier, E. and Monteil, J.-M. (1995) ‘Where does self-
esteem come from? The inXuence of private vs. public individuation’,
Social Behavior and Personality, 23, pp. 70–81.
Humphreys, J. (1995) ‘Gender eVects on children’s performance on the
Tower of Hanoi as a manual and a computer task’, Unpublished disser-
tation, Department of Psychology, University of Southampton.
Hutchins, G., Hall, W. and Colbourn, C. (1993) ‘Patterns of students’
interactions with hypermedia systems’, Interacting with Computers, 5, pp.
295–313.
Inhelder, B. and Piaget, J. (1958) The Growth of Logical Thinking from
Childhood to Adolescence. New York: Basic Books.
Jackson, A., Fletcher, B. and Messer, D. (1986) ‘A survey of microcomputer
use and provision in primary schools’, Journal of Computer Assisted Learn-
ing, 2, pp. 45–55.
Janssen Reinen, I. and Plomp, T. (1997) ‘Information technology and gender
equality: A contradiction in terms?’, Computers and Education, 28 (2), pp.
65–78.
Järvelä, S. (1995) ‘The cognitive apprenticeship model in a technologically
rich learning environment: Interpreting the learning interaction’, Learn-
ing and Instruction, 5 (3), pp. 237–59.
Johnson, D. and Johnson, R. (1986) ‘Computer assisted cooperative learn-
ing’, Educational Technology, January, 12–18.
Joiner, R., Messer, D., Light, P. and Littleton, K. (1994) ‘Talking and
learning: A preliminary investigation into productive interaction’ in
References 107
Foot, H., Howe, C., Anderson, A. Tolmie, A. and Warden, D. (eds.)

Group and Interactive Learning, Southampton, Computational Mechan-
ics Publications.
Joiner, R., Messer, D., Steele, F., Light, P. and Littleton, K. (1993) ‘Peer
presence, gender and computer experience in computer based problem
solving’, Paper presented at the ESRC InTER Seminar on Collaborative
Learning, Oxford, February.
Karau, S. and Williams, K. (1993) ‘Social loaWng: A meta-analytic review
and theoretical integration’, Journal of Personality and Social Psychology,
29, pp. 135–74.
Kirkup, G. (1992) ‘The social construction of computers’ in Kirkup, G. and
Keller, S. (eds.) Inventing Women: Science, Gender and Technology,
Oxford, Polity Press.
Kliman, M. (1985) ‘A new approach to infant and early primary Mathemat-
ics’, DAI Research Paper 241, Edinburgh: Department of ArtiWcial
Intelligence, University of Edinburgh.
Kluger, A. and DeNisi, A. (1996) ‘The eVects of feedback interventions on
performance: A historical review, a meta-analysis and a preliminary
feedback intervention theory’, Psychological Bulletin, 119, pp. 254–84.
Kutnick, P. and Rogers, C. (eds.) (1994) Groups in Schools, London, Cassell.
Lave, J. and Wenger, E. (1991) Situated Learning: Legitimate Peripheral Par-
ticipation, Cambridge, Cambridge University Press.
Lee, M. (1993) ‘Gender, group composition and peer interaction in com-
puter based co-operative learning’, Journal of Educational Computing
Research, 4, pp. 549–77.
Light, P. (1979) The Development of Social Sensitivity, Cambridge, Cambridge
University Press.
Light, P. (1986) ‘Context, conservation and conversation’ in Richards, M.
and Light, P. (eds.) Children of Social Worlds, Cambridge, Polity Press.
Light, P. (1994) ‘Peer interaction and peer comparison as facilitators of
children’s learning’, Invited Plenary Address, First International Confer-
ence on Group and Interactive Learning, Glasgow, September.
Light, P. (1997) ‘Computers for learning: Psychological perspectives’, Jour-
nal of Child Psychology and Psychiatry, 38 (5), pp. 497–504.
Light, P., Buckingham, N. and Robbins, A. (1979) ‘The conservation task as
an interactional setting’, British Journal of Educational Psychology, 49, pp.
304–10.
Light, P. and Colbourn, C. (1987) ‘The role of social processes in children’s
microcomputer use’ in Kent, W. and Lewis, R. (eds.) Computer Assisted
Learning in the Social Sciences and Humanities, Oxford, Blackwell.
Light, P. and Foot, T. (1987) ‘Peer interaction and microcomputer use’
Rassegna di Psicologia, 4, pp. 93–104.
Light, P., Foot, T., Colbourn, C. and McClelland, I. (1987) ‘Collaborative
interactions at the microcomputer keyboard’, Educational Psychology, 7,
pp. 13–21.
Light, P. and Gilmour, A. (1983) ‘Conservation or conversation? Contextual
facilitation of inappropriate conservation judgements’, Journal of Experi-
108 References
mental Child Psychology, 36, pp. 356–63.

Light, P. and Glachan, M. (1985) ‘Peer interaction and problem solving’,
Educational Psychology, 5, pp. 217–25.
Light, P., Gorsuch, C. and Newman, J. (1987) ‘Why do you ask? Context
and communication in the conservation task’, European Journal of Psy-
chology of Education, 2, pp. 73–82.
Light, P. and Light, V. (1999) ‘Reaching for the sky: Computer supported
tutorial interaction in a conventional university setting’ in Littleton, K.
and Light, P. (eds.) Learning with Computers: Analysing productive interac-
tion, London, Routledge.
Light, P., Littleton, K., Bale, S., Messer, D. and Joiner, R. (in press) ‘Gender
and social comparison eVects in computer based problem solving’,
Learning and Instruction.
Light, P., Littleton, K., Messer, D. and Joiner, R. (1994) ‘Social and com-
municative processes in computer based learning’, European Journal of
Psychology of Education, 9, pp. 93–109.
Light, P. and Nix, C. (1983) ‘‘‘Own view’’ versus ‘‘good view’’ in a perspec-
tive-taking task’ Child Development, 54, pp. 480–83.
Light, P. and Perret-Clermont, A-N. (1991) ‘Social context eVects in learn-
ing and testing’ in Light, P., Sheldon, S. and Woodhead, M. (eds.)
Learning to Think, London, Routledge.
Light, P., Sheldon, S. and Woodhead, M. (eds.) (1991) Learning to Think,
London, Routledge.
Linnakylä, P. (1996) ‘High expectations – High achievement on literacy’,
Paper presented at the 1996 World Conference on Literacy , Philadel-
phia, March 12–15.
Littleton, K. (1995) ‘Children and Computers’ in Bancroft, D. and Carr, R.
(eds.) InXuencing Children’s Development, Oxford, Blackwell.
Littleton, K. (1996) ‘Girls and information technology’ in Murphy, P. and
Gipps, C. (eds.) Equity in the classroom: Towards an EVective Pedagogy for
Girls and Boys, London, Falmer Press/UNESCO Publishing.
Littleton, K., Ashman, H., Light, P., Artis, J., Roberts, T. and Oosterwegel,
A. (1999) ‘Gender, task contexts and children’s performance on a com-
puter-based task’, European Journal of Psychology of Education.
Littleton, K. and Bannert, M. (in press) ‘Gender and IT: Contextualising
diVerence’ in Bliss, J., Light, P. and Säljö, R. (eds.) Learning Sites: Social
and Technological Contexts for Learning, Oxford: Pergamon.
Littleton, K., Light, P., Joiner, R., Messer, D. and Barnes, P. (1992) ‘Pairing
and gender eVects on children’s computer-based learning’, European
Journal of Psychology of Education, 7, pp. 311–24.
Littleton, K., Light, P., Joiner, R., Messer, D. and Barnes, P. (1998) ‘Gen-
der, task scenarios and children’s computer-based problem solving’,
Educational Psychology, 18 (3), pp. 327–40.
Littleton, K., Light, P., Robertson, A. and Beeton, A. (in preparation) ‘A task
by any other name . . . ? Academic attainment, task contexts, and
children’s on-task performance’.
Lockheed, M., Nielson, A. and Stone, M. (1985) ‘Determinants of micro-
References 109
computer literacy in high school students’, Journal of Educational Com-

puting Research, 1, pp. 81–96.
McGarrigle, J. and Donaldson, M. (1975) ‘Conservation accidents’, Cogni-
tion, 3, pp. 341–50.
Manstead, A. and Semin, G. (1980) ‘Social facilitation eVects: Mere enhan-
cement of dominant response?’, British Journal of Social and Clinical
Psychology, 19, pp. 119–36.
Marshall, H. and Weinstein, R. (1984) ‘Classroom factors aVecting students’
self evaluations: An interactional model’, Review of Educational Research,
54, pp. 301–25.
Martin, R. (1991) ‘School children’s attitudes towards computers as a func-
tion of gender, course subjects and availability of home computers’,
Journal of Computer Assisted Learning, 7, pp. 187–94.
Mash, E. and Hedley, J. (1975) ‘EVect of observer as a function of prior
history of social interaction’, Perceptual and Motor Skills, 40, pp. 659–69.
Malone, T. (1981) ‘Toward a theory of intrinsically motivating instruction’,
Cognitive Science, 4, pp. 333–69.
Mercer, N. (1992) ‘Culture, context and the construction of knowledge in
the classroom’ in Light, P. and Butterworth, G. (eds.) Context and
Cognition: Ways of Learning and Knowing, Hemel Hempstead, Herts,
Harvester Wheatsheaf.
Mercer, N. (1995) The Guided Construction of Knowledge, Clevedon, Avon,
Multilingual Matters.
Mercer, N. and Fisher, E. (1992) ‘How do teachers help children to learn?
An analysis of teachers’ interventions in computer-based activities’,
Learning and Instruction , 2, pp. 339–55.
Mercer, N. and Wegerif, R. (1999) ‘Is exploratory talk productive talk?’ in
Littleton, K. and Light, P. Learning with Computers: Analysing Productive
Interaction, London, Routledge.
Mevarech, Z., Silber, O. and Fine, D. (1991) ‘Learning with computers in
small groups’, Journal of Educational Computing Research, 7, pp. 233–43.
Miller, S. (1982) ‘On the generalisability of conservation’, British Journal of
Miller, S. and Brownell, C. (1975) ‘Peers, persuasion and Piaget’, Child
Development, 46, pp. 992–97.
Monteil, J.-M. (1988) ‘Comparaison sociale: Strategies individuelle et medi-
ations socio-cognitives’, European Journal of Psychology of Education, 3,
pp. 3–18.
Monteil, J.-M. and Huguet, P. (1991) ‘Insertion sociale, categorisation so-
ciale et activites cognitives’, Psychologie Francaise, 36, pp. 35–46.
Monteil, J.-M. and Huguet, P. (in press) The Social Psychology of Cognition,
Hove, Erlbaum.
Mugny, G. and Doise, W. (1978) ‘Socio-cognitive conXict and structures of
individual and collective performances’, European Journal of Social Psy-
chology, 8, pp. 181–92.
Mugny, G., Perret-Clermont and Doise, A.-N. (1981) ‘Interpersonal co-
ordinations and sociological diVerences in the construction of the intel-
110 References
lect’ in Stevenson, G. and Davis, G. (eds.) Applied Social Psychology,

vol. I, Chichester, Wiley.
Murphy, P. (1993) ‘Assessment and gender’ in Bourne, J. (ed.) Thinking
Through Primary Practice, London, Routledge.
Murray, J. (1974) ‘Social learning and cognitive development’, British Jour-
nal of Psychology, 65, pp. 151–60.
Nathan, M. and Resnick, L. (1994) ‘Less can be more: Unintelligent tutoring
based on psychological theories and experimentation’ in Vosniadou, S.,
De Corte E. and Mandl, H. (eds.) Technology Based Learning Environ-
ments, Berlin, Springer-Verlag.
Newman, D., GriYn, P. and Cole, M. (1989) The Construction Zone: Working
for Cognitive Change in Schools, Cambridge, Cambridge University Press.
Newton P. and Beck E. (1993) ‘Computing: An ideal occupation for
women?’ in Beynon, J. and MacKay, H. (eds.) Computers into Classrooms:
More Questions than Answers, London, Falmer Press.
Niemiec, R. and Walberg, H. (1987) ‘Comparative eVects of computer
assisted instruction: A synthesis of reviews’, Journal of Educational Com-
puting Research, 3, pp. 19–37.
Packer, M. (1993) ‘Away from internalization’ in Forman, E., Minick, N.
and Stone, A. (1993) Contexts for Learning: Sociocultural Dynamics in
Children’s Development, Oxford, Oxford University Press.
Papert, S. (1980) Mindstorms: Children, Computers and Powerful Ideas, New
York, Harvester Wheatsheaf.
Papert, S. (1994) The Children’s Machine, London, Harvester Wheatsheaf.
Pea, R. and Kurland, M. (1984) ‘The cognitive eVects of learning computer
programming’, New Ideas in Psychology, 2, pp. 137–68.
Pelgrum W. and Plomp, T. (1993) The IEA Study of Computers in Education,
Oxford, Pergamon.
Perret-Clermont, A.-N. (1980) Social Interaction and Cognitive Development in
Children, London, Academic Press.
Pheasey, K. and Underwood, G. (1994) ‘Collaboration and discourse during
computer based problem solving’ in Foot, H., Howe, C., Anderson, A.,
Tolmie, A. and Warden, D. (eds.) Group and Interactive Learning, South-
ampton, Computational Mechanics Publications.
Phelps, E. and Damon, W. (1989) ‘Problem solving with equals: Peer collab-
oration as a context for learning mathematical and spatial concepts’,
Journal of Educational Psychology, 81, pp. 639–46.
Piaget, J. (1932) The Moral Judgement of the Child, London, Routledge.
Piaget, J. and Inhelder, B. (1956) The Child’s Conception of Space, London,
Routledge.
Pozzi, S., Healy, L. and Hoyles, C. (1993) ‘Learning and interaction in
groups with computers: When do ability and gender matter?’, Social
Development, 2, pp. 222–41.
Prout, A. (1998) Concluding remarks, Conference on Children and Social
Exclusion, Centre for the Social Study of Childhood, Hull University.
Provenzo, E. (1991) Video Kids: Making Sense of Nintendo, Cambridge, MA:
Harvard University Press.
References 111
Radziszewska, B. and RogoV, B. (1988) ‘InXuence of adult and peer collab-

oration on children’s planning skills’, Developmental Psychology, 24, pp.
840–48.
Resnick, L., Pontecorvo, C. and Säljö, R. (1997) ‘Discourse, tools and
reasoning’ in Resnick, L., Säljö, R., Pontecorvo, C. and Burge, B. (eds.)
Discourse, Tools and Reasoning: Essays on Situated Cognition, Berlin and
New York, Springer-Verlag.
Ringelmann, M. (1913) ‘Recherches sur les moteurs animes: Travail de
l’homme’, Annales de l’Institut Nationale Agronomique, 12, pp. 1–40.
Roazzi, A. and Bryant, P. (1992) ‘Social class, context and cognitive develop-
ment’ in Light, P. and Butterworth, G. (eds.) Context and Cognition:
Ways of Learning and Knowing, Hemel Hempstead, Harvester Wheat-
sheaf.
Robertson, S. I., Calder, J., Fung, P., Jones, A. and O’Shea, T. (1995)
‘Computer attitudes in an English secondary school’, Computers and
Education, 24, pp. 3–81.
Robinson-Staveley, K. and Cooper, J. (1990) ‘Mere presence, gender
and reaction to computers: Studying human–computer interaction in
the social context’, Journal of Experimental Social Psychology, 26,
pp. 168–83.
RogoV. B. (1990) Apprenticeship in Thinking: Cognitive Development in Social
Context, Oxford, Oxford University Press.
RogoV, B. and Gardner, W.P. (1984) ‘Guidance in cognitive development:
An examination of mother–child instruction’, in RogoV, B and Lave, J.
(eds.) Everyday Cognition: Its Development in Social Context, Cambridge,
MA, Harvard University Press.
Rommetveit, R. (1992) ‘Outlines of a dialogically based social-cognitive
approach to human cognition and communication’ in Wold, A. (ed.) The
Dialogical Alternative: Towards a Theory of Language and Mind, Oslo,
Scandinavian Press.
Roschelle, J. and Teasley, S. (1995) ‘The construction of shared knowledge
in collaborative problem solving’ in O’Malley, C. (ed.) Computer Sup-
ported Collaborative Learning, Berlin, Springer Verlag.
Rose, N. (1990) ‘Psychology as a ‘‘social’’ science’ in Parker, I. and Shotter,
J. (eds.) Deconstructing Social Psychology, London, Routledge.
Rosenthal, T. and Zimmerman, B. (1972) ‘Modelling by exempliWcation and
instruction in training conservation’, Developmental Psychology, 6, pp.
392–401.
Ruble, D. (1983) ‘The development of social comparison processes and their
role in achievement-related self-socialisation’ in Higgins, E., Ruble, D.
and Hartup, W. (eds.) Social Cognition and Social Development: A Socio-
Cultural Perspective, Cambridge, Cambridge University Press.
Ruble, D. and Flett, G. (1988) ‘ConXicting goals in self-evaluative informa-
tion seeking: Developmental and ability level analyses’, Child Develop-
ment, 59, pp. 97–106.
Ruble, D. and Frey, K. (1991) ‘Changing patterns of comparative behaviour
as skills are acquired: A functional model of self-evaluation’ in Suls, J.
112 References
and Wills, T. (eds.) Social Comparison: Contemporary Theory and Re-

search, Hillsdale, NJ., Erlbaum.
Russell, J. (1981) ‘Why socio-cognitive conXict may be impossible’, Educa-
tional Psychology, 1, pp. 159–69.
Russell, J. (1982) ‘Propositional attitudes’ in Beveridge, M. (ed.) Children
Thinking Through Language, London, Edward Arnold.
Säljö, R. (1995) ‘Mental and physical artefacts in cognitive practices’ in
Reimann, P. and Spada, H. (eds.) Learning in Humans and Machines:
Towards an Interdisciplinary Learning Science, Oxford, Pergamon.
Säljö, R. (1997) ‘Heavenly talk. Discourse, artifacts and children’s under-
standing of elementary astronomy’, Seminar presentation. Milton Key-
nes, Open University, November.
Sanna, L. (1992) ‘Self eYcacy theory: Implications for social facilitation and
social loaWng’, Journal of Personality and Social Psychology, 62, pp. 774–
86.
Scanlon, E., IssroV, K. and Murphy, P. (1999) ‘Collaborations in a primary
classroom’ in Littleton, K. and Light, P. (eds.) Learning with Computers:
Analysing Productive Interaction, London, Routledge.
Schank, R. and Cleary, C. (1995) Engines for Education, Hillsdale, NJ, Er-
lbaum.
SchoWeld, J. (1995) Computers and Classroom Culture, Cambridge, Cambridge
University Press.
Scribner, S. (1975) ‘Situating the experiment in cross cultural research’ in
Riegel, K.F. and Meacham J. A. (eds.) The Developing Individual in a
Changing World: Historical and Cultural Issues, The Hague, Mouton.
Scribner, S. and Cole, M. (1981) The Psychology of Literacy, Cambridge, MA,
Harvard University Press.
Seta, J. and Hassan, R. (1980) ‘Awareness of prior success and failure: A
critical factor in task performance’, Journal of Personality and Social
Shashaani, L. (1993) ‘Gender based diVerence in attitudes towards com-
puters’, Computers and Education, 20 (2), pp. 169–81.
Sheingold, K., Hawkins, J. and Char, C. (1984) ‘‘‘I’m the thinkist, you’re
the typist’’: The interaction of technology and the social life of class-
rooms’, Technical Report 27, Bank Street College of Education, New
York.
Shrock, S. and Stepp, S. (1991) ‘The role of the child micro-computer expert
in an elementary classroom: A theme emerging from a naturalistic study’,
Journal of Research on Computing in Education, 23 (4), pp. 545–59.
Siann, G., Durndell, A., McLeod, H. and Glissov, P. (1988) ‘Stereotyping in
relation to gender gap in participation in computing’, Educational Re-
search, 30, pp. 98–103.
Siegler, R. (1976) ‘Three aspects of cognitive development’, Cognitive Psy-
chology, 8, pp. 481–520.
Silverman, I. and Geiringer, E. (1973) ‘Dyadic interaction and conservation
induction’, Child Development, 44, pp. 815–20.
Skinner, B. F. (1965) ‘ReXections on a decade of teaching machines’ in
References 113
Glaser, R. (ed.) Teaching Machines and Programmed Learning, Washin-

gton, National Education Association.
Smith, R. (1991) ‘A prototype futuristic technology for distance education’
in Scanlon, E. and O’Shea, T. (eds.) New Directions in Educational
Technology, Berlin, Springer-Verlag.
Snow, R. (1994) ‘Abilities in academic tasks’ in Sternberg, R. and Wagner,
R. (eds.) Mind in Context: Interactionist Perspectives on Human Intelligence,
Cambridge, Cambridge University Press.
Straker, A. (1989) Children Using Computers, Oxford, Blackwell.
Stutz, E. (1996) ‘Is electronic entertainment hindering children’s play and
social development?’ in Gill, T. (ed.) Electronic Children: How Children
are Responding to the Information Revolution, London, National Children’s
Bureau.
Subrahmanyam, K. and GreenWeld, P. (1994) ‘EVect of video-game practice
on spatial skills in girls and boys’, Journal of Applied Developmental
Suchman, L. (1987) Plans and Situated Actions, Cambridge, Cambridge
University Press.
Suppes, P. (1966) ‘The use of computers in education’, ScientiWc American,
215, pp. 207–20.
Swann, J. (1997) ‘Tinkertown: Reading and re-reading children’s talk around
the computer’ in Wegerif, R. and Scrimshaw, P. (eds.) Computers and
Talk in the Primary Classroom, Clevedon, Avon, Multilingual Matters.
Tharp, R. and Gallimore, R. (1988) Rousing Minds to Life: Teaching, Learning
and Schooling in Social Context, Cambridge, Cambridge University Press.
Tizard, B., Blatchford, P., Burke, J., Farquhar, C. and Plewis, I. (1988)
Young Children at School in the Inner City, Hove, Erlbaum.
Todman, J. and Dick, G. (1993) ‘Primary children and teachers’ attitudes to
computers’, Computers in Education, 20, pp. 199–203.
Todman, J. and File, P. (1990) ‘A scale for children’s attitudes to com-
puters’, School Psychology International, 11, 71–75.
Tomasello, M., Kruger, A. and Ratner, H. (1993) ‘Cultural learning’, Behav-
ioral and Brain Sciences, 16, pp. 495–552.
Topping, K. (1994) ‘Peer tutoring’ in Kutnick, P. and Rogers, C. (eds.)
Groups in Schools, London, Cassell.
Triplett, N. (1898) ‘The dynamogenic factors in pacemaking and competi-
tion’ American Journal of Psychology, 9, pp. 507–33.
Turkle, S. and Papert, S. (1990) ‘Epistemological pluralism: Styles and
voices within the computer culture’, Signs: Journal of Women in Culture
and Society, 16, pp. 128–57.
Underwood, G., Jindal, N. and Underwood, J. (1994) ‘Gender diVerences
and eVects of cooperation in a computer based language task’, Education-
al Research, 36, pp. 63–74.
Underwood, G., McCaVrey, M. and Underwood, J. (1990) ‘Gender diVeren-
ces in a computer based language task’ Educational Research, 32, pp. 44–9.
Underwood, J. and Underwood, G. (1999) ‘Task eVects on co-operative and
collaborative learning with computers’ in Littleton, K. and Light, P.
114 References
(eds.) Learning with Computers: Analysing Productive Interaction, London,

Routledge.
Valsiner, J. (1997) ‘Bounded indeterminacy in discourse processes’ in Coll,
C. and Edwards, D. (eds.) Teaching, Learning and Classroom Discourse,
Madrid, Fundación Infancia y Aprendizaje.
Vygotsky, L.S. (1978) Mind in Society: The Development of Higher Psychologi-
cal Processes, Cambridge, MA, Harvard University Press.
Vygotsky, L.S. (1990) ‘Genesis of the higher mental functions’ in Light, P.,
Sheldon, S. and Woodhead, M. (eds.) Learning to Think, London,
Routledge.
Watson, M. (1997) ‘Improving group work at the computer’ in Wegerif, R.
and Scrimshaw, P. (eds.) Computers and Talk in the Primary Classroom,
Clevedon, Avon, Multilingual Matters.
Webb, N., Ender, P. and Lewis, S. (1986) ‘Problem solving strategies and
group processes in small groups learning computer programming’,
American Educational Research Journal, 23, pp. 243–61.
Wegerif, R. and Mercer, N. (1997) ‘A dialogical framework for researching
peer talk’ in Wegerif, R. and Scrimshaw, P. (eds.) Computers and Talk in
the Primary Classroom, Clevedon, Avon, Multilingual Matters.
Wertsch, J. (1985) Vygotsky and the Social Formation of Mind, Cambridge,
MA, Harvard University Press.
Wertsch, J. (1997) ‘The socio-cultural approach to learning’, Paper pres-
ented to an Inaugural Conference of the Centre for Learning in Or-
ganisations, School of Education, University of Bristol, January.
Wertsch, J. (1998) Mind as Action, Oxford, Oxford University Press.
Wertsch, J., Tulviste, P. and Hagstrom, F. (1993) ‘A socio-cultural approach
to agency’ in Forman, E., Minick, N.and Stone, A. (1993) Contexts for
learning: Sociocultural Dynamics in Children’s Development, Oxford, Ox-
ford University Press.
White, S. (1996) ‘Foreword’ in Cole, M. Cultural Psychology: A Once and
Future Discipline, Cambridge, MA, The Belknap Press of Harvard Uni-
versity Press.
Whitelock, D., O’Shea, T., Taylor, J., Scanlon, E., Clark P. and O’Malley,
C. (1993) ‘The inXuence of peer interaction and peer presence in com-
puter supported learning’, Paper presented to CAL 93 Conference ‘CAL
into the Mainstream’, York, April.
Whitley, B. (1997) ‘Gender diVerences in computer-related attitudes and
behaviour: A meta-analysis’, Computers in Human Behaviour, 13 (1), pp.
1–22.
Wilder, G., Mackie, D. and Cooper, J. (1985) ‘Gender and computers: Two
surveys of computer related attitudes’, Sex Roles, 13, pp. 215–28.
Williams, J., Harkins, S. and Latane, B. (1981) ‘IdentiWability as a deterrent
to social loaWng’, Journal of Personality and Social Psychology, 40, pp.
303–11.
Wood, D. (1986) ‘Aspects of teaching and learning’. In Richards, M and
Light, P. (eds.) Children of Social Worlds, Cambridge, Polity Press.
Wood, D., Bruner, J. and Ross, G. (1976) ‘The role of tutoring in problem
References 115
solving’, Journal of Child Psychology and Psychiatry, 17, pp. 89–100.

Wood, D. and Wood, H. (1996) ‘Vygotsky, tutoring and learning’, Oxford
Review of Education, 22, pp. 5–16.
Wood, D. and Wood, H. (in press) ‘Contingency in tutoring and learning’,
Learning and Instruction.
Woodhead, M. (1999) ‘Reconstructing developmental psychology – Some
Wrst steps’, Children and Society, 13 (1), 1–17.
Woodhead, M, Faulkner, D. and Littleton, K. (eds.) (1998) Cultural Worlds
of Early Childhood, London, Routledge.
Woodhead, M. Faulkner, D. and Littleton, K. (eds.) (1999) Making Sense of
Social Development, London, Routledge.
Zajonc, R. (1965) ‘Social facilitation’, Science, 149, pp. 269–74.
Index
Compiled by Judith Lavender
academic disciplines, impact on and gender, 54–5

performance, 78–81 Honeybears studies, 45–9, 50–1, 59
aVordance, 98 King and Crown studies, 36–9, 40–1,
age, impact on problem solving, 25 42, 50
agency, 93 Mastermind studies, 20–2, 25
appropriation of cultural tools, 12 mediating role of computer, 96–7
art, performance in context of, 78–81 overview of, 25–6, 91–2
artefacts, 96 symmetry in, 45–8
see also computers; tools talk in see verbal interaction
associationist learning theory, 27–8 theories of, xiv–xv
attunement, 98 Tower of Hanoi studies, 14–20, 25
see also groupwork; mixed-sex groups;
balance beam studies, 22–4, 25–6 peer interaction; same-sex groups
Bruner, J., 9 computer games
Buckingham, N., 5–6 context of, 84–9
Burnham, W., 74 macho culture of, 53
computer programming, 29, 31
CAI (computer-assisted instruction), 27, computer software
28, 31 contrast studies and gender, 60–4, 70–2
Cazden, C., 11 gender stereotypes in, 53, 58–60, 71–2
Charbonnier, E., 77 computer tasks
children, viewpoints of, 97–9 balance beam studies, 22–4, 25–6
coaching development of, 32–3
in Tower of Hanoi studies, 17 game versus test contexts, 84–9
see also instruction Honeybears studies, 43–51, 58–9, 60–4,
coactive working, 65, 66, 67, 68, 69 65–6, 67, 70
see also peer presence King and Crown studies, 33–42, 50,
code-breaking problems see Mastermind 56–8, 60–1, 63–4
studies Mastermind studies, 20–2, 25
cognition Pirates studies, 61–4
development as social process, xvi–xvii Princesses study, 67
impact of peer interaction on Tower of Hanoi studies, 18–20, 25,
development, 2–5, 9, 10–11 83–4
interactive approaches to, 92–3 computer tests, context of, 84–9
social-cognition approach, xiii–xiv computer-assisted instruction (CAI), 27,
see also thinking 28, 31
cognitive conXict see socio-cognitive computers
conXict associationist approach to, 27–8
collaborative learning collaborative approach to, 30–1
balance beam studies, 22–4, 25–6 constructionist approach to, 29–30
with computers, 30–1 context of use, 81–9, 97
cultural context of, 93–6 in education, xv–xvi, 27–31
116
Index 117
and gender: attitudes, 52, 66–7, 69–70, Gabrenya, W., 77
71; context, 81–9; performance, 56, Gallimore, R., 8, 9
58, 59, 61, 62, 65–6, 69–71; styles, games
53–4, 65–6; usage, 53, 88 gendered performance in context of,
group use of see collaborative learning 84–9
growing interest in, 26, 27 macho culture of, 53
mediating role of, 96–7 gender
concrete operations tasks, 3–4 attitudes to computers, 52, 66–7, 69–70,
conXict see socio-cognitive conXict 71, 88–9
conservation judgements peer interaction in learning, 54–5
Perret-Clermont’s work on, 4–5 performance with computers, 56, 58, 59,
problems of standard tests on, 5–6 61, 62, 65–6, 69–71; contextual
constructive approach, 2–5, 29–30 inXuences, 81–9
context software contrast studies, 60–4, 70–2
of collaboration, 93–6 stereotypes in software, 53, 58–60, 71–2
computer and manual, 81–4 style of computer use, 53–4, 65–6
disciplinary, 78–81 usage of computers, 53, 88
of performance, 97–9 see also mixed-sex groups; same-sex
contingent strategies, 9–10 groups
Cooper, J., 71 Gorsuch, C., 6
Cottrell, N., 75 groupwork
cultural context of collaboration, 93–6 move towards, xv
cultural learning, theories of, xiv–xv and use of computers, xv–xvi
cultural psychology, xvii see also collaborative learning; mixed-sex
see also socio-cultural approach groups; same-sex groups
cultural tools, 11–12
Hall, J., 71
decentration, 2–3 Hassan, R., 76
developmental psychology Hedley, J., 76
nature of, 99–100 history of experience, 93
social comparison in, 73 home computers, gendered use of, 53
disciplines, performance in context of, Honeybears studies
78–81 collaborative work on, 45–9, 50–1, 59
Doise, W., xvi, 3–4 features of, 43–4
Donaldson, M., 5 gender inXuences in, 58–9; peer
Durkin, K., 73 interaction, 65–6, 67, 70; software
contrasts, 60–4
education impact of peer presence, 49–50, 51
increasing use of computers in, xv–xvi, HuV, C., 71
27; role of psychology in, 27–31 Huguet, P., 74, 75–7, 78
educational software Humphreys, J., 83–4
contrast studies and gender, 60–4,
70–2 identity construction, link with thinking,
gender stereotypes in, 53, 58–60, 71–2 99
Edwards, D., 10 individual working see coactive working;
experimental research peer presence
context of, 95–6 instruction
see also computer tasks computer-assisted, 27, 28, 31
intelligent tutoring systems, 28
feedback, impact on performance, 74, in Tower of Hanoi studies, 16–17
76–7, 78 intellectual development, as social process,
Festinger, L., 73 xvi–xvii
Flett, G., 73 intelligent tutoring systems, 28
Forman, E., 11 interaction see peer interaction; social
friendship, impact on productivity, 24–5 interaction
118 Index
Joiner, R., 71 see also collaborative learning; mixed-sex
justiWcation, in balance beam studies, 23, groups; same-sex groups
24, 25–6 peer presence
impact on performance, 49–50, 51, 65,
King and Crown studies 74–6
collaborative work on, 36–9, 40–1, 42, see also coactive working
50 peer tutors, 10
features of, 33–6, 40 performance
gender inXuences in, 56–8; software impact of computer context on, 81–4
contrasts, 60–1, 63–4 impact of discipline context on, 78–81
knowledge, social construction of, 92 impact of friendship on, 24–5
impact of peer presence on, 49–50, 51,
Latane, B., 77 74–6
learning impact of social comparison on, 74–8
changing notions of, 92 relationship with gender, 56, 58, 59, 61,
contingent strategies in, 9–10 62, 65–6, 69–71
dimensions of, 99 as situated, 97–9
and social comparison, 77–84 Perret-Clermont, A.-N., 4–5, 7
Vygotskian approach to interaction in, Piaget, J., xvi, 2–3
8–11 Pirates studies, 61–4
see also collaborative learning Princesses study, 67
Light, P., 5–6, 71 productivity see performance
Littleton, K., 71 programming computers, 29, 31
Logo, 29, 31 psychology of learning, impact on
computer use, 27–31
Malone, T., 89
Manstead, A., 75 Resnick, L., 92, 93
Marshall, H., 74 Ringelmann, M., 76
Mash, E., 76 Rittle, R., 75
Mastermind studies, 20–2, 25 Robbins, A., 5–6
maths, performance in context of, 78–81 Robinson-Staveley, K., 71
mediated action, 11–12 role division, in King and Crown studies,
memory tests, gender and context of, 39
78–83 Roschelle, J., 47
Mercer, N., 10 Ross, G., 9
Messer, D., 71 Ruble, D., 73
mixed-sex groups working with computers, Russell, J., 5
54–5
Honeybears studies, 59 same-sex groups working with computers,
interaction and coaction, 64–70, 71 54, 55
King and Crown studies, 56–8 Honeybears studies, 59
modelling, 1, 17, 28 interaction and coaction, 64–70, 71
Monteil, J.-M., 74, 75–8 King and Crown studies, 56–8
mouse control, and gender, 58, 59 Sanna, L., 76–7
scaVolding, 9, 10
Newman, J., 6 school, girls’ understanding of, 87–8
Sekerak, G., 75
operational modes of thought, 2–4 Semin, G., 75
Seta, J., 76
Papert, S., 29 simulated worlds, 29–30
peer interaction single-sex groups see same-sex groups
and gender in learning, 54–5 situated learning approach, xvii
impact on development, 2–5, 9, 10–11 see also context
impact of friendship on, 24–5 Skinner, B. F., 27
on social class diVerences, 7 social class diVerences, 7
Index 119
social comparison, 73–7 Tharp, R., 8, 9
and children’s learning, 77–84 thinking
social construction link with identity construction, 99
of knowledge, 92 see also cognition
see also constructivist approach tools, 11–12, 96
social interaction see also computers
Vygotskian approach to learning and, Tower of Hanoi studies, 14–20, 25, 83–4
8–11 Triplett, N., 74
see also peer interaction tutoring see instruction
social learning theory, 1–2
social loaWng, 76–7 unstructured interaction, 17, 18
social-cognition approach, xiii–xiv
socio-cognitive conXict, 4, 17, 26, 91 verbal interaction, 54, 58
reservations about, 5–7 in Honeybears studies, 48
socio-cultural approach, 11–12 in King and Crown studies, 40–1
software in Mastermind studies, 21–2, 25
contrast studies and gender, 60–4, 70–2 in Tower of Hanoi studies, 19–20,
gender stereotypes in, 53, 58–60, 71–2 25
Steele, F., 71 verbal justiWcation, in balance beam
Stroop test, 75 studies, 23, 24, 25–6
structured interaction, 17, 18, 19 village task, 3–4
style of computer use, 53–4, 65–6 Vygotsky, L. S., xvi–xvii, 8–9, 11
symbolic tools, 11–12 inXuence of, 7–8, 9–11
symmetry, in collaborative working, 45–8
Wack, D., 75
talk see verbal interaction Wang, Y., 77
teaching Weinstein, R., 74
contingent strategies in, 9–10 Wertsch, J., 11–12
ZPD in theory of, 8 Wood, D., 9–10
Teasley, S., 47
test, gendered performance in context of, Zajonc, R., 74–5
84–9 zone of proximal development (ZPD), 8–9

Social Processes in Children's Learning: Paul Light and Karen Littleton

Uploaded by

Copyright:

Available Formats

You might also like

Social Processes in Children's Learning: Paul Light and Karen Littleton

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Social Processes in Children's Learning: Paul Light and Karen Littleton

Uploaded by

Copyright:

Available Formats

Social Processes in

Paul Light and Karen Littleton

Cambridge University Press

This page intentionally left blank

This book is about children’s learning and problem-solving behaviour.

paul li gh t is Pro-Vice-Chancellor of Bournemouth University, and

k ar e n l i t tl e to n currently holds an ESRC personal research fellow-

Titles published in the series

2. Learning to Read and Write: A Cross-Linguistic Perspective

3. Children’s Understanding of Biology and Health

This page intentionally left blank

Paul Light and Karen Littleton

© Paul Light and Karen Littleton 1999

First published in printed format 1999

A catalogue record for the original printed book is available

ISBN 0 511 01202 0 virtual (netLibrary Edition)

This page intentionally left blank

List of Wgures page x

1 Peer interaction and learning: perspectives and starting points 1

2.1 The Tower of Hanoi page 15

This page intentionally left blank

Two decades ago, in a monograph on ‘The Development of Social

social cognition as concerned with ‘cognition as a product of social interac-

from the standpoint of the individual developing learner or of the social

long emphasised the potential of computers to support eVective learning

approach to understanding how peer interaction can facilitate learning

Figure 4.5 is reprinted with permission from Carfax Publishing Limited.

This page intentionally left blank

Modelling success: the social learning theory approach

evidence that children who initially performed at ‘pre-operational’ levels

Construction of understanding through

would create a pressure towards resolution of diVerences: ‘Criticism is

potential beneWts of peer interaction for the development of children’s

Reservations about socio-cognitive conXict

Learning as the co-construction of understanding

‘Vygotskian’ approaches in relation to a whole range of research problems

between the actual developmental level as determined by individual

In further studies Wood and colleagues showed that the adoption of a

A socio-cultural approach to learning

do mental arithmetic is just as much a cultural tool as is a pocket

Piaget and Vygotsky: complementary constructions

these perspectives have served their purpose in framing this Weld of

The Wrst Tower of Hanoi study

Figure 2.1 The Tower of Hanoi

The second Tower of Hanoi study

The third Tower of Hanoi study

Figure 2.2 The computer version of the Tower of Hanoi

The Wrst Mastermind study

The second Mastermind study

interaction between pre- to post-test improvement and the level of argu-

Two balance beam studies

Figure 2.3 The balance beam task

children’s justiWcations, but these were always received without com-

Friendship and productivity in peer interaction

anxiety. There were no signiWcant diVerences in terms of expressed

Computers and instruction

through small incremental steps. The principal mechanism for achieving

Computers and construction